Help Implementing Virtualized Servers in a small but steadily growing business
Tags:
-
Business Computing
-
Servers
Last response: in Business Computing
ropern01
July 11, 2014 8:44:07 AM
Hello everyone, I would just like to start by saying thank you very much to anyone who is willing to offer input. I have been reading up on this topic for about three months trying to educate myself and I feel like I have developed some paralysis from analysis. I really need to get these guys upgraded as they are now running on one mission critical server. I don't want to invest another dime in that old server so I've got to get a new server out quick before lightning strikes. Also I want to apologize for starting off asking a book of questions. I am just trying to be thorough with my thoughts and needs so that I am not missing anything or off base with my ideas. If anyone is willing to help but wants to just cut to the chase, my biggest concerns are listed in the list of concerns section.
Mission and client info
I am hoping for a little guidance with implementing virtualized servers in a small business. I would like to use this as a base model for future clients also. This client is a manufacturing company with Two servers and about 25 workstations. This does not include phones and Ipads which mostly just run over wireless to get internet and email. Their email is hosted so no problems there. This client desperately needs new servers and honestly could stand to upgrade the infrastructure. I am trying to keep the budget around 10K but can probably push it higher if I can give good enough reason with some hard information to back it up. I am going to totally re-create the domain so not really worried about decommissioning the old servers. Right now there are about twenty five workstations. The main app is a relatively small SQL database running on sql2005. They also have a Synology 412+ configured with 4 2TB Western Digital Red drives configured for RAID 10. This drive is also backed up to an external drive plugged in to the NAS and to IDrive online. This is where they host their main parts files for design.
My Plan
Let me start with my choice of hyper visor. Because the IT needs of this business are relatively small I have decided to go with Hyper-V because I think it will be cheaper with the licensing. So my plan is to have two physical servers on site running. Let's call them server 1 and server 2. Server 1 will be the main server. It will actually run two virtual servers to start with and have the capability to add at least a couple more if necessary. This is the one that I recommend buying new with a service level agreement in case of a hardware failure. Server 2 will be simply for replicating server 1 and available on standby in case of a hardware failure with server 1. Server 2 does not have to be as powerful as server 1. It really doesn't even have to be brand new but It has to be at least powerful enough to run the virtual servers at an acceptable level in case of a hardware failure in server 1. This setup will allow us to operate without worrying about being down for any extended period because of a hardware failure. It also almost covers you on disaster recovery. Mix in online backup and you are close. Below I just kind of listed the components necessary to make the upgrade happen. Pls let me know if I am missing anything or if I don't have a handle on any of this.
Also I already have quotes from Dell if anyone wants to let me know if I am getting appropriate hardware.
Server 1
New server with a good service level agreement on hardware. I don't recommend this but If money is tight we could cut corners on the redundancy features since it would have a service level agreement and server 2 will be available for failover. Again my philosophy is too many redundancy/backup options are still not enough.
Server 2
Recommend new low end server since(theoretically) it is only going to have to run for a day or so in case of emergency. We could also utilize it for more than just disaster recovery if it is a decent server. Again If too tight we can find a good used server for cheap that will basically sit there and be a replication/insurance box.
2 UPS Battery Backup
I recommend battery backups for both servers.
2 Server 2012 license
Have to buy a license for both boxes, good news is that server 2012 allows for 2 virtual licenses
SQL Server
I believe that we can use the 2005 version that is on our server right now. Unfortunately, If not the license for sql is expensive and I will look into other options such as my SQL. Testing testing testing.
Workstation CAL
Have to buy at least a 25 seat CAL so that windows workstations can connect to the server
Gigabit switch
Will speed the whole network up. Most of the existing wiring is CAT 5E so we could make the jump to gigabit speed.
Firewall
Will assist in defending the network and will speed up inbound/outbound wan traffic. Current ninja turtle is an fvs318....ssssllllllllooooooooowwwwww
List of concerns
1. Should I try to achieve HA and if so should I go the smb over say a company like Starwind of possibly even over the NAS, however I don't think that is the best choice from what I have read. I just don't think we have the budget to shell out for a SAN, nor do I really like the idea of one piece of hardware that could fail. Even with redundancy things can go wrong, like say what if it lost ability to receive power or the array controller puked?
2. Which hardware route to take concerning the servers. Load balance with two decent servers or 1 loaded server and 1 entry level machine to use only in case of hardware failure? I realize that having a computer just sit there is wasting resources but two mediocre servers will never give
3. Which components do I really need to make a virtual box run smoothly. I think I understand it to be memory, processor cores, processor strength and array controller followed by decent cache, a speedy bus finally out of a set of nicely balanced nics. Wouldn't It would be nice to have someone write an app that would figure out safe operating specs for you like a loan calculator. It would measure the I/O through put vs. processor load vs. memory use and give you a minimum requirement per vm. OK, it's 3a.m. and I have officially gone delirious.
4. Hard drives selection and configuration. Should I go with a couple of 15k SAS drives or ssd mirrored for the OS? What about for the data section? raid 10? Not sure if the virtual factor makes a difference when making this selection. Have heard to enable jumbo frames. Also I noticed the big manufacturers drives are outrageous but I am wondering how buying your own would affect the SLA.
5. Minimum memory to use. Again thinking this will only two nodes but I am thinking 32GB if I go the way of the Beast and the beater. If I use two decent servers maybe 16 each? IDK that seems a little low to me.
6. I think that we will be ok running SQL on these virtual machines. We already purchased SQL 05 but not sure how it's going to do with 2012 in a virtual environment. It should be fine but you know how these things can go. Looks like we will be doing some testing before implementing these. Anyone have any experience with this
7. Licensing. I hope I am understanding what I believe to be our licensing costs. So two copies of server 2012R2 covers us for four virtual servers running virtual server. I wanted to ask a licensing professional if we were not using the backup server, (it is just sitting there powered off unless we have to boot it up in which the other server is off) do we have to buy a seperate license for it? However I read if you are doing replication then you have to have a seperately licensed target. Anyway then there is the CAL licenses for the workstations.
Recommendations
Please feel free to recommend any business model, product or service that you know/had a good experience with relative to the category. My goal is to keep the budget at a minimum all while giving my client maximum value to meet their business needs.
1. A more solid computer\network model for a small but steadily growing business relative to the budget. Which is hopefully $10-$15k.
2. Server/Server hardware/server method
3. Gigabit Switch
4. Firewall
5. Server Antivirus
6. Replication method
7. Anything that will make life as an IT consultant easier. (take off three weeks and quit!)
Again I want to give a big THANKS to everyone and anyone willing to take the time to help with anything!!!
Mission and client info
I am hoping for a little guidance with implementing virtualized servers in a small business. I would like to use this as a base model for future clients also. This client is a manufacturing company with Two servers and about 25 workstations. This does not include phones and Ipads which mostly just run over wireless to get internet and email. Their email is hosted so no problems there. This client desperately needs new servers and honestly could stand to upgrade the infrastructure. I am trying to keep the budget around 10K but can probably push it higher if I can give good enough reason with some hard information to back it up. I am going to totally re-create the domain so not really worried about decommissioning the old servers. Right now there are about twenty five workstations. The main app is a relatively small SQL database running on sql2005. They also have a Synology 412+ configured with 4 2TB Western Digital Red drives configured for RAID 10. This drive is also backed up to an external drive plugged in to the NAS and to IDrive online. This is where they host their main parts files for design.
My Plan
Let me start with my choice of hyper visor. Because the IT needs of this business are relatively small I have decided to go with Hyper-V because I think it will be cheaper with the licensing. So my plan is to have two physical servers on site running. Let's call them server 1 and server 2. Server 1 will be the main server. It will actually run two virtual servers to start with and have the capability to add at least a couple more if necessary. This is the one that I recommend buying new with a service level agreement in case of a hardware failure. Server 2 will be simply for replicating server 1 and available on standby in case of a hardware failure with server 1. Server 2 does not have to be as powerful as server 1. It really doesn't even have to be brand new but It has to be at least powerful enough to run the virtual servers at an acceptable level in case of a hardware failure in server 1. This setup will allow us to operate without worrying about being down for any extended period because of a hardware failure. It also almost covers you on disaster recovery. Mix in online backup and you are close. Below I just kind of listed the components necessary to make the upgrade happen. Pls let me know if I am missing anything or if I don't have a handle on any of this.
Also I already have quotes from Dell if anyone wants to let me know if I am getting appropriate hardware.
Server 1
New server with a good service level agreement on hardware. I don't recommend this but If money is tight we could cut corners on the redundancy features since it would have a service level agreement and server 2 will be available for failover. Again my philosophy is too many redundancy/backup options are still not enough.
Server 2
Recommend new low end server since(theoretically) it is only going to have to run for a day or so in case of emergency. We could also utilize it for more than just disaster recovery if it is a decent server. Again If too tight we can find a good used server for cheap that will basically sit there and be a replication/insurance box.
2 UPS Battery Backup
I recommend battery backups for both servers.
2 Server 2012 license
Have to buy a license for both boxes, good news is that server 2012 allows for 2 virtual licenses
SQL Server
I believe that we can use the 2005 version that is on our server right now. Unfortunately, If not the license for sql is expensive and I will look into other options such as my SQL. Testing testing testing.
Workstation CAL
Have to buy at least a 25 seat CAL so that windows workstations can connect to the server
Gigabit switch
Will speed the whole network up. Most of the existing wiring is CAT 5E so we could make the jump to gigabit speed.
Firewall
Will assist in defending the network and will speed up inbound/outbound wan traffic. Current ninja turtle is an fvs318....ssssllllllllooooooooowwwwww
List of concerns
1. Should I try to achieve HA and if so should I go the smb over say a company like Starwind of possibly even over the NAS, however I don't think that is the best choice from what I have read. I just don't think we have the budget to shell out for a SAN, nor do I really like the idea of one piece of hardware that could fail. Even with redundancy things can go wrong, like say what if it lost ability to receive power or the array controller puked?
2. Which hardware route to take concerning the servers. Load balance with two decent servers or 1 loaded server and 1 entry level machine to use only in case of hardware failure? I realize that having a computer just sit there is wasting resources but two mediocre servers will never give
3. Which components do I really need to make a virtual box run smoothly. I think I understand it to be memory, processor cores, processor strength and array controller followed by decent cache, a speedy bus finally out of a set of nicely balanced nics. Wouldn't It would be nice to have someone write an app that would figure out safe operating specs for you like a loan calculator. It would measure the I/O through put vs. processor load vs. memory use and give you a minimum requirement per vm. OK, it's 3a.m. and I have officially gone delirious.
4. Hard drives selection and configuration. Should I go with a couple of 15k SAS drives or ssd mirrored for the OS? What about for the data section? raid 10? Not sure if the virtual factor makes a difference when making this selection. Have heard to enable jumbo frames. Also I noticed the big manufacturers drives are outrageous but I am wondering how buying your own would affect the SLA.
5. Minimum memory to use. Again thinking this will only two nodes but I am thinking 32GB if I go the way of the Beast and the beater. If I use two decent servers maybe 16 each? IDK that seems a little low to me.
6. I think that we will be ok running SQL on these virtual machines. We already purchased SQL 05 but not sure how it's going to do with 2012 in a virtual environment. It should be fine but you know how these things can go. Looks like we will be doing some testing before implementing these. Anyone have any experience with this
7. Licensing. I hope I am understanding what I believe to be our licensing costs. So two copies of server 2012R2 covers us for four virtual servers running virtual server. I wanted to ask a licensing professional if we were not using the backup server, (it is just sitting there powered off unless we have to boot it up in which the other server is off) do we have to buy a seperate license for it? However I read if you are doing replication then you have to have a seperately licensed target. Anyway then there is the CAL licenses for the workstations.
Recommendations
Please feel free to recommend any business model, product or service that you know/had a good experience with relative to the category. My goal is to keep the budget at a minimum all while giving my client maximum value to meet their business needs.
1. A more solid computer\network model for a small but steadily growing business relative to the budget. Which is hopefully $10-$15k.
2. Server/Server hardware/server method
3. Gigabit Switch
4. Firewall
5. Server Antivirus
6. Replication method
7. Anything that will make life as an IT consultant easier. (take off three weeks and quit!)
Again I want to give a big THANKS to everyone and anyone willing to take the time to help with anything!!!
More about : implementing virtualized servers small steadily growing business
-
Reply to ropern01
ropern01
July 12, 2014 7:43:00 PM
Best solution
First off, bravo. You have done your research and you have a good idea of where to begin, but you also aren't afraid to ask for assistance in the areas you know you have questions. Your post has given good details about what you are wanting to accomplish along with clearly stating what questions you have, and being open to receiving input and suggestions. Great beginning!
So, on to the questions. This type of environment is similar to what I work with as well at several offices. Most of those offices choose to stick with a single server simply because of cost. However, I have one customer who is setting up an almost identical environment. Just like you are looking at, I would recommend going with with two servers so you can leverage Replication within Hyper-V. In reality it wouldn't make much difference if you decided to use one more powerful server and one less powerful for replication, but to me that seems a little counter-productive. It is more efficient to split your workload to two identical servers. They don't both have to be uber powerful, as again the only situation you may need to run ALL of your VMs on a single host is in the event of a planned or unplanned outage and you have to migrate workloads. However, don't undercut yourself on performance capabilities either as you are bound to grow. You will more than likely always add VMs or demand to your server, very rarely will it decrease.
So, it's hard really for me to say exactly what kind of performance capabilities you may need, mainly because of your database. Domain controllers and file servers generally don't need much horsepower to run. The system I just set up currently has one of the two Hyper-V hosts installed. They went with an HP ProLiant ML360p G8 server with a single E5-2620 V2 six-core processor and 32 GB of RAM. They don't come near using that amount of RAM yet, but it's cheaper to order it to start then trying to add later down the road often. On that host I currently have four VMs set up in Hyper-V. A primary and a secondary domain controller (the secondary will be migrated to the second server once it is online), a file server, and an application server. They also currently have two older VMs running a terminal server and a database system on another older server with VMWare that we will eventually convert over and move to the new servers.
Looking some more at server hardware, I'd ask how much data storage you are needing? If you don't have a ton of space requirement for shared files, it might be best to move that over to your new servers. The benefit here is you can virtualize your storage server and perhaps leverage faster and more robust hardware. You also get data duplication in the form of replication to your secondary server. The down side is you have to have storage capacity on BOTH servers to support your file server needs. If you do move the file server to your main servers, you can utilize the Synology NAS as your data store for backing up your servers. This is also what I have done at a few locations.
Many places recommend that you set up a couple small capacity hard drives in RAID 1 just for the OS of your Hyper-V server. I do this usually because it allows me to change the storage array or capacity of the rest of my "datastore" for my Hyper-V data without actually affecting the base OS installation or partition at all. However, just as many people anymore will suggest otherwise. If you want throughput and storage capacity instead of using up two of your available hard drives for your OS, put in a RAID 10 array and just carve out a small partition for your OS. This definitely makes sense as you get better capacity and throughput this way. Again, I don't know for sure how much total capacity you need but going with a RAID 10 array is probably going to be the best bet there. You can go with SSDs for part of your datastore, such as storing a small but high-demand database, but be aware that not all SSDs are created equal, and many will not hold up for too long to the demanding stress of a server environment. Compatibility and function of different SSDs will vary depending upon your controller, but basically the more expensive SSD drives out there are going to be what you want to look at. Still, if capacity is more of an issue, looking at SAS hard drives will probably be more cost-effective, but of course won't offer the performance of SSDs.
Your network plays a critical role in how effective your server systems will be as well. Take for example this same customer I was mentioning before. Previously their old server was running two demanding VMs, a terminal server plus a database server. Even though their server system could handle the load fine, they had ALL of the networking running through a single gigabit interface on the server, which was daisy chained through two desktop switches, one of them a 10/100 switch! Obviously there were issues!
You don't have to go huge to get a decent and stable network in place. I like the HP ProCurve line of switches because you get good quality, good support, and a cheap price. The main thing here is what features do you need? Is there a need for setting up multiple VLANs? Layer 3 routing? Quality of Service features? If no, then you should be fine going with something like the ProCurve 1810 series switches. A single 48-port managed gigabit switch runs around $600 which isn't bad! If you want some more management features or prefer the Cisco brand, I've recently worked with some SG200 and SG500X series switches and they are very nice as well. You're going to be paying slightly more for the Cisco stuff, but it has more management features and capabilities than the ProCurve 1810 series has. We ended up setting up some SG500X switches because they support 10 GbE for future expandability as well as stacking. If you have few enough connections you can get by with a single 48-port switch, but you may consider getting a 24-port switch for your "core" and a 48-Port for the "access" for the rest of your network. Your core switch will connect your primary infrastructure (firewall, servers, and other switches) and your access switch connects all your end computers, printers, etc. This will also give you more room to grow, as if you add more servers you connect more devices to your core switch, and if you add more computers you connect them to your access switch.
Along with this you should be able to set up NIC teaming through Hyper-V to leverage greater throughput. A single gigabit NIC will give you about 100 MB/s of throughput, which a standard 7k SATA hard drive will saturate doing a large file transfer. Obviously if you are using a RAID 10 array of high speed SAS drives or even SSDs then you are going to be bottlenecked by that gigabit connection. NIC teaming allows you to leverage multiple ethernet adapters for greater overall throughput. Your switch doesn't even need to support LACP if you are using Windows Server 2012 for your NIC teaming, but it is recommended. For example in our setup we put all four of the gigabit interfaces into a single NIC team with LACP and Address Hash. Likewise in the switch we configured all four interfaces there in a LAG (Link Aggregation Group) with LACP and no standby adapters. We can push a solid 300 MB/s through from server to server and that is actually the maximum of our storage array currently, so there's room for more if we add more hard drives! We also can throw in 10 GbE adapters connected directly between the servers for high speed network for the replication and migration data.
When it comes to your firewall, there are a lot of options here. If you want something cheap but powerful then the Ubiquiti EdgeRouter line is quite nice. The management interface has matured quite a bit since I last used one, but it still may be a bit of a learning curve or be more difficult to get up and running with some features that are very easy to manage with something like the Sonicwall TZ series. I've used Sonicwalls at many offices and they are very simple to manage. The big thing here is to find something that is easy for you to work with, and is powerful enough for the throughput you need. Are you going to be doing VPN connections? VoIP? What is your internet connection speed? The other option here is going with an open-source system like Untangle. I haven't worked much with Untangle yet, but what I have used of it I really do like it. There's a lot of free packages that give you tons of features and if you have a spare computer sitting around that supports it you can have something up and going with all the features and capabilities of the expensive firewalls but with little or no cost. It is possible to even virtualize your Untangled system onto server hardware to move it around if needed, though I've personally not tried that out so can't make recommendations either way.
There's tons more of course to talk about, all sorts of things to consider, but for now I'll leave it at that and see if you have other information or questions from there! Hopefully this helps some, and I hope others may post their input here as well!
So, on to the questions. This type of environment is similar to what I work with as well at several offices. Most of those offices choose to stick with a single server simply because of cost. However, I have one customer who is setting up an almost identical environment. Just like you are looking at, I would recommend going with with two servers so you can leverage Replication within Hyper-V. In reality it wouldn't make much difference if you decided to use one more powerful server and one less powerful for replication, but to me that seems a little counter-productive. It is more efficient to split your workload to two identical servers. They don't both have to be uber powerful, as again the only situation you may need to run ALL of your VMs on a single host is in the event of a planned or unplanned outage and you have to migrate workloads. However, don't undercut yourself on performance capabilities either as you are bound to grow. You will more than likely always add VMs or demand to your server, very rarely will it decrease.
So, it's hard really for me to say exactly what kind of performance capabilities you may need, mainly because of your database. Domain controllers and file servers generally don't need much horsepower to run. The system I just set up currently has one of the two Hyper-V hosts installed. They went with an HP ProLiant ML360p G8 server with a single E5-2620 V2 six-core processor and 32 GB of RAM. They don't come near using that amount of RAM yet, but it's cheaper to order it to start then trying to add later down the road often. On that host I currently have four VMs set up in Hyper-V. A primary and a secondary domain controller (the secondary will be migrated to the second server once it is online), a file server, and an application server. They also currently have two older VMs running a terminal server and a database system on another older server with VMWare that we will eventually convert over and move to the new servers.
Looking some more at server hardware, I'd ask how much data storage you are needing? If you don't have a ton of space requirement for shared files, it might be best to move that over to your new servers. The benefit here is you can virtualize your storage server and perhaps leverage faster and more robust hardware. You also get data duplication in the form of replication to your secondary server. The down side is you have to have storage capacity on BOTH servers to support your file server needs. If you do move the file server to your main servers, you can utilize the Synology NAS as your data store for backing up your servers. This is also what I have done at a few locations.
Many places recommend that you set up a couple small capacity hard drives in RAID 1 just for the OS of your Hyper-V server. I do this usually because it allows me to change the storage array or capacity of the rest of my "datastore" for my Hyper-V data without actually affecting the base OS installation or partition at all. However, just as many people anymore will suggest otherwise. If you want throughput and storage capacity instead of using up two of your available hard drives for your OS, put in a RAID 10 array and just carve out a small partition for your OS. This definitely makes sense as you get better capacity and throughput this way. Again, I don't know for sure how much total capacity you need but going with a RAID 10 array is probably going to be the best bet there. You can go with SSDs for part of your datastore, such as storing a small but high-demand database, but be aware that not all SSDs are created equal, and many will not hold up for too long to the demanding stress of a server environment. Compatibility and function of different SSDs will vary depending upon your controller, but basically the more expensive SSD drives out there are going to be what you want to look at. Still, if capacity is more of an issue, looking at SAS hard drives will probably be more cost-effective, but of course won't offer the performance of SSDs.
Your network plays a critical role in how effective your server systems will be as well. Take for example this same customer I was mentioning before. Previously their old server was running two demanding VMs, a terminal server plus a database server. Even though their server system could handle the load fine, they had ALL of the networking running through a single gigabit interface on the server, which was daisy chained through two desktop switches, one of them a 10/100 switch! Obviously there were issues!
You don't have to go huge to get a decent and stable network in place. I like the HP ProCurve line of switches because you get good quality, good support, and a cheap price. The main thing here is what features do you need? Is there a need for setting up multiple VLANs? Layer 3 routing? Quality of Service features? If no, then you should be fine going with something like the ProCurve 1810 series switches. A single 48-port managed gigabit switch runs around $600 which isn't bad! If you want some more management features or prefer the Cisco brand, I've recently worked with some SG200 and SG500X series switches and they are very nice as well. You're going to be paying slightly more for the Cisco stuff, but it has more management features and capabilities than the ProCurve 1810 series has. We ended up setting up some SG500X switches because they support 10 GbE for future expandability as well as stacking. If you have few enough connections you can get by with a single 48-port switch, but you may consider getting a 24-port switch for your "core" and a 48-Port for the "access" for the rest of your network. Your core switch will connect your primary infrastructure (firewall, servers, and other switches) and your access switch connects all your end computers, printers, etc. This will also give you more room to grow, as if you add more servers you connect more devices to your core switch, and if you add more computers you connect them to your access switch.
Along with this you should be able to set up NIC teaming through Hyper-V to leverage greater throughput. A single gigabit NIC will give you about 100 MB/s of throughput, which a standard 7k SATA hard drive will saturate doing a large file transfer. Obviously if you are using a RAID 10 array of high speed SAS drives or even SSDs then you are going to be bottlenecked by that gigabit connection. NIC teaming allows you to leverage multiple ethernet adapters for greater overall throughput. Your switch doesn't even need to support LACP if you are using Windows Server 2012 for your NIC teaming, but it is recommended. For example in our setup we put all four of the gigabit interfaces into a single NIC team with LACP and Address Hash. Likewise in the switch we configured all four interfaces there in a LAG (Link Aggregation Group) with LACP and no standby adapters. We can push a solid 300 MB/s through from server to server and that is actually the maximum of our storage array currently, so there's room for more if we add more hard drives! We also can throw in 10 GbE adapters connected directly between the servers for high speed network for the replication and migration data.
When it comes to your firewall, there are a lot of options here. If you want something cheap but powerful then the Ubiquiti EdgeRouter line is quite nice. The management interface has matured quite a bit since I last used one, but it still may be a bit of a learning curve or be more difficult to get up and running with some features that are very easy to manage with something like the Sonicwall TZ series. I've used Sonicwalls at many offices and they are very simple to manage. The big thing here is to find something that is easy for you to work with, and is powerful enough for the throughput you need. Are you going to be doing VPN connections? VoIP? What is your internet connection speed? The other option here is going with an open-source system like Untangle. I haven't worked much with Untangle yet, but what I have used of it I really do like it. There's a lot of free packages that give you tons of features and if you have a spare computer sitting around that supports it you can have something up and going with all the features and capabilities of the expensive firewalls but with little or no cost. It is possible to even virtualize your Untangled system onto server hardware to move it around if needed, though I've personally not tried that out so can't make recommendations either way.
There's tons more of course to talk about, all sorts of things to consider, but for now I'll leave it at that and see if you have other information or questions from there! Hopefully this helps some, and I hope others may post their input here as well!
-
Reply to choucove
Share
Related resources
- Help us Build a Checklist for a Virtualized Business Server Implementation - Forum
- Need help with choosing a small business server. - Forum
- Help deciding on a small business server build - Forum
- Help Building a Small Business Server - Forum
- Windows Server 2008, Hyper V virtualization for small business - Forum
ropern01
July 13, 2014 7:34:45 PM
The more I think about it the more I agree with you that I think two decent servers load balanced is the better way to go. I still think that I may beef one up a tad more than the other to try make sure that most of the time I will have good speed on the SQL box and the other box should be plenty powerful enough to handle a domain controller. Idea is in case of the power box going down we could borrow some of it's memory until the power box was fixed! Anyway...I'm going to say they will be a lot closer to each other than I originally and will both be used leveraging load balancing.
As far as what server to buy. I have a couple of quotes from dell I listed below. I was going to go with them simply because I am more familiar with their servers. I know HP makes a good machine as well. Definitely not opposed to using their equipment. I googled the ml360p g8 and had trouble finding them. Not sure if they are just rare or it's a typo. Found a bunch of dl360p g8 and ml350p g8. That being said from what I could tell it seems similar to the R320's with the single processor. Looks like hp is using a little better processor family via cpuboss.com. Guess point being if they are similar and you are running your vm with relative ease that sort of lets me know I should have a similar experience provided the infrastructure is up to par. Also not sure how crazy I should go with the array controller but since it plays a big part in the path to I/O I guess I better get to reading. So really I am feeling like the second quote I got from Dell with a 420 and a 320 is getting close to what I want. These quotes include a switch supposedly at half price but I think it will be overkill since we have basically a flat network. So I need to figure out which array controllers and which drives I can afford.
https://onedrive.live.com/redir?resid=D1F2FE79A1AD8366!413&authkey=!AHlxQ3wp5u4Ln9c&ithint=folder%2c
I am still not clear though about setting replication. I am thinking there are really two ways to go about it.
1. Replication. Simply setup a target vm to replicate to on one the second server.
2. HA. Setup some sort of SAN for storage pooling. Then both servers look to that to run their VM's. Also it can be done similarly but cheaper using smb 3 protocol. Same thing as the SAN just setup can now be ran from a NAS or Service IE Starwind that utilizes smb.
Is that correct or am I on dog food?
I am thinking I like what you said about the Raid 10 and then partition for the OS. We have only used up about 75 gigs of crucial data over 10 years so I don't feel like we need a ton of storage. I really like the samsung 840 pro ssd's but not sure I can squeeze that in the budget nor am I sure if they are server grade. Four 512 drives would be about 2k. Plus I am a little gun shy...lol...Just experienced an OCZ going bad on a vip workstation and what a pain. It acted just like bad memory. Then to top it off I had issues with the efi bios and recognizing partitions, acronis kept converting fat to gpt and It kept crashing. So I think you have helped me decide on another choice. Raid 10 and then I still may look in to the ssd option but I am leaning toward sas drives hopefully 15k! Should be fine hopefully since the machines are being replicated anyway. One more thing, I guess that utilizing a server for storage instead of the synology will depend on where we are with the price in relation to the budget but again I don't think we need a ton of space so I will definitely consider that.
I think that the procurve would be perfect. I was actually looking at those a while back. Like I say the network is flat and I can't see a real advantage to subnet it off with it being as small as it is. So no need to go with high end switch. As far as the 48 port, I think I like the idea of two 24 ports over one 48. That way if we have a hardware failure we aren't completely down.
I will definitely look into the Ubiquiti EdgeRouter line. Thanks again for the recommendation, I'm sure it's just what the Dr ordered. I am familiar with the sonicwall's but im not sure they wanna pickup another bill with the subscription.
We don't have VoIP but I do have a VPN setup. Believe it or not I actually configured the Synology's VPN app which works great with the default client that comes with windows. The only other thing that I am thinking is that we have a DVR with 8 feeds but that comes back to the infrastructure. I want to try and implement 10GB anywhere I can so that eventually we might get to that. As of now all of the cabling is cat 5e so we are bottlenecked at GB there.
I probably missed some things because this is a lot of stuff to cover, especially with the phone ringing, txts every 5 minutes, kids running around crazy, wife hitting me with honey do list..lol. What I didn't miss is to say thanks again for taking your time to help with my problems. I appreciate it more than you know.
So to summarize
Hardware highlights
Two similar servers server 1 and server 2
Dell R420 and Dell R320
May only put 1 processor in 420 to start
at least 6 cores
32GB on R420 and 16GB on R320
Redundant power supplies (may cut to one if budget gets busted)
Array controller R420 H710 and R320 H310
Server 1
VM 1 SQL Server
VM 2 PDC replicated
Server 2
VM 3 PDC
VM 4 SQL Replication
Questions
1. Licensing. I am gonna purchase Microsoft Server 2012R2 Standard with each server.
I believe that I am allowed two VM's per Hypervisor. So that means I should be able to intall Server 2012 on to two vm's each server. I am wondering if I were to use veeam or something to continually backup say my domain controller into a file instead of replicating, would I be able to use the other license(the one I was going to use to replicate both servers to ) for something else like a BDC and whatever we may need. Just a thought
2. The question I asked earlier. Are there basically two ways to configure this setup?
I am still not clear though about setting replication. I am thinking there are really two ways to go about it.
A. Replication. Simply setup a target vm to replicate to on one the second server.
B. HA. Setup some sort of SAN for storage pooling. Then both servers look to that to run their VM's. Also it can be done similarly but cheaper using smb 3 protocol. Same thing as the SAN just setup can now be ran from a NAS or Service IE Starwind that utilizes smb.
3. What to use for backup. I have seen where there is a script written in power shell that a guy has posted. He says it is all you need to backup VM's but I have to cover their rears on this. In my mind this is more of in case you get a corrupt registry or get a virus, since ultimately your replication is like a backup.
4. Anything I am leaving out here.
As far as what server to buy. I have a couple of quotes from dell I listed below. I was going to go with them simply because I am more familiar with their servers. I know HP makes a good machine as well. Definitely not opposed to using their equipment. I googled the ml360p g8 and had trouble finding them. Not sure if they are just rare or it's a typo. Found a bunch of dl360p g8 and ml350p g8. That being said from what I could tell it seems similar to the R320's with the single processor. Looks like hp is using a little better processor family via cpuboss.com. Guess point being if they are similar and you are running your vm with relative ease that sort of lets me know I should have a similar experience provided the infrastructure is up to par. Also not sure how crazy I should go with the array controller but since it plays a big part in the path to I/O I guess I better get to reading. So really I am feeling like the second quote I got from Dell with a 420 and a 320 is getting close to what I want. These quotes include a switch supposedly at half price but I think it will be overkill since we have basically a flat network. So I need to figure out which array controllers and which drives I can afford.
https://onedrive.live.com/redir?resid=D1F2FE79A1AD8366!413&authkey=!AHlxQ3wp5u4Ln9c&ithint=folder%2c
I am still not clear though about setting replication. I am thinking there are really two ways to go about it.
1. Replication. Simply setup a target vm to replicate to on one the second server.
2. HA. Setup some sort of SAN for storage pooling. Then both servers look to that to run their VM's. Also it can be done similarly but cheaper using smb 3 protocol. Same thing as the SAN just setup can now be ran from a NAS or Service IE Starwind that utilizes smb.
Is that correct or am I on dog food?
I am thinking I like what you said about the Raid 10 and then partition for the OS. We have only used up about 75 gigs of crucial data over 10 years so I don't feel like we need a ton of storage. I really like the samsung 840 pro ssd's but not sure I can squeeze that in the budget nor am I sure if they are server grade. Four 512 drives would be about 2k. Plus I am a little gun shy...lol...Just experienced an OCZ going bad on a vip workstation and what a pain. It acted just like bad memory. Then to top it off I had issues with the efi bios and recognizing partitions, acronis kept converting fat to gpt and It kept crashing. So I think you have helped me decide on another choice. Raid 10 and then I still may look in to the ssd option but I am leaning toward sas drives hopefully 15k! Should be fine hopefully since the machines are being replicated anyway. One more thing, I guess that utilizing a server for storage instead of the synology will depend on where we are with the price in relation to the budget but again I don't think we need a ton of space so I will definitely consider that.
I think that the procurve would be perfect. I was actually looking at those a while back. Like I say the network is flat and I can't see a real advantage to subnet it off with it being as small as it is. So no need to go with high end switch. As far as the 48 port, I think I like the idea of two 24 ports over one 48. That way if we have a hardware failure we aren't completely down.
I will definitely look into the Ubiquiti EdgeRouter line. Thanks again for the recommendation, I'm sure it's just what the Dr ordered. I am familiar with the sonicwall's but im not sure they wanna pickup another bill with the subscription.
We don't have VoIP but I do have a VPN setup. Believe it or not I actually configured the Synology's VPN app which works great with the default client that comes with windows. The only other thing that I am thinking is that we have a DVR with 8 feeds but that comes back to the infrastructure. I want to try and implement 10GB anywhere I can so that eventually we might get to that. As of now all of the cabling is cat 5e so we are bottlenecked at GB there.
I probably missed some things because this is a lot of stuff to cover, especially with the phone ringing, txts every 5 minutes, kids running around crazy, wife hitting me with honey do list..lol. What I didn't miss is to say thanks again for taking your time to help with my problems. I appreciate it more than you know.
So to summarize
Hardware highlights
Two similar servers server 1 and server 2
Dell R420 and Dell R320
May only put 1 processor in 420 to start
at least 6 cores
32GB on R420 and 16GB on R320
Redundant power supplies (may cut to one if budget gets busted)
Array controller R420 H710 and R320 H310
Server 1
VM 1 SQL Server
VM 2 PDC replicated
Server 2
VM 3 PDC
VM 4 SQL Replication
Questions
1. Licensing. I am gonna purchase Microsoft Server 2012R2 Standard with each server.
I believe that I am allowed two VM's per Hypervisor. So that means I should be able to intall Server 2012 on to two vm's each server. I am wondering if I were to use veeam or something to continually backup say my domain controller into a file instead of replicating, would I be able to use the other license(the one I was going to use to replicate both servers to ) for something else like a BDC and whatever we may need. Just a thought
2. The question I asked earlier. Are there basically two ways to configure this setup?
I am still not clear though about setting replication. I am thinking there are really two ways to go about it.
A. Replication. Simply setup a target vm to replicate to on one the second server.
B. HA. Setup some sort of SAN for storage pooling. Then both servers look to that to run their VM's. Also it can be done similarly but cheaper using smb 3 protocol. Same thing as the SAN just setup can now be ran from a NAS or Service IE Starwind that utilizes smb.
3. What to use for backup. I have seen where there is a script written in power shell that a guy has posted. He says it is all you need to backup VM's but I have to cover their rears on this. In my mind this is more of in case you get a corrupt registry or get a virus, since ultimately your replication is like a backup.
4. Anything I am leaving out here.
-
Reply to ropern01
m
0
l
So let's talk a little bit about High Availability (HA) and what is now infamously known as the inverted pyramid of doom.
High availability can be a huge benefit for any business who wants to minimize downtime as much as possible. If you look at your company and start crunching the numbers and listing out the services and people who are affected if your servers go down, it can be quite shocking to see how critical of a role they play in daily business life anymore. This is why businesses with 50 or more employees who are looking at new servers may be considering HA. With this situation, you would set up a SAN device of some sort (either a server, SAN storage device, or scale-out file server cluster) which would actually hold the data for your virtual machines. This SAN network is accessible from multiple host servers but not always simultaneously. So, if you have Server1 running your SQL database and suddenly Server1 has a power supply fail and shuts off, the cluster would detect this failure and start up the virtual machine for your SQL database again on your Server2 system meaning very short downtime and little to no data loss.
That being said, HA is definitely not something to jump into just because of its benefits because it's caveats can be even more critical. For example, what happens if your SAN server suffers a power supply failure, motherboard failure, RAID controller failure, corrupted partition, etc. etc.? The heart of your system is really that SAN and if it goes down you have suffered a cataclysmic failure, also known as the inverted pyramid of doom. SANs can be unbelievably expensive. There are many types out there, but you're looking at a minimum of $10k - $15k just to get a basic SAN system going. That leaves you no budget for the necessary network infrastructure, server hosts, or licensing. SAN and cluster environments are also more complex to manage, configure, and maintain overall.
Replication is the cheap man's version of HA. It's making duplicates of the data and changes to that data at regular intervals to another server. In the event that the primary server goes down, the virtual machine has to be manually started up on the other server system, so it's not automatic. There can also be several minutes worth of lost time or data, as it will go off the last good replicated datapoint when the server starts up again. However, setting up Replication only has cost in the actual host servers, there is no need for extra SAN storage and configuration.
One thing I'm not completely positive on, but from what I understand is replicated VMs do not require an additional license and will work off the original VM's Windows Server 2012 license so long as you aren't running both simultaneously of course. So in the situation that you are talking about above, you'd have Server1 with a Windows Server 2012 license, and Server2 with a Windows Server 2012 license, and this would actually give you a total of four VMs, two on each server. That means you should be able to set up Server1 with SQL server, a backup domain controller, and Server2 with your primary domain controller and perhaps an application server or something else as needed.
When it comes to replication, it can be pretty forgiving about varied hardware. For instance, for a live migration the servers have to have the same type of processors (Intel or AMD) and it is recommended that they have the same features or are the same platform (for instance, LGA2011.) However, they don't have to be identical. From the information and research that I have read on it, however, it is recommended to keep the servers as close to identical as possible (RAID controllers, NICs, processors, etc.) This eliminates compatibilities moving the VMs between different underlying hardware that may require different driver support, etc. I also recommend keeping them the same processor platform (socket type, chipset, etc.) because of physical hardware support. If you need to buy additional RAM the same RAM is compatible in both servers that way. If you need to upgrade processors they are the same compatible processors.
When it comes to backup again there are numerous ways and products to do so. I've tried out the free version of VEEAM and it worked pretty well but didn't have any of the automatic features I was looking for without purchasing a full version of the software which none of my customers were for given the cost. Mostly what I use is a two-fold backup system. First, within the VM. For example with a file server virtual machine, I utilize a data copy application like Uranium Backup (which is free) to schedule automatic backup of all the actual business data (documents, pictures, etc.) to a network shared location or additional external hard drive. There are a TON of options here within Uranium, such as doing daily backups for a full week, then overwriting the backups day by day the next week. This gives you a full week's worth of previous versions to go back to if necessary but obviously increases the amount of storage capacity you need.
This is my critical data backup. With this backup I have all the original files in their original individual formats and not in some proprietary compressed file. So, in the event that everything went haywire and I had to get access to some files within minutes for someone, I can unplug the external hard drive, plug it into any other computers, and I have access to all of the company data. Ideally this should be set up in a rotation. For instance, have one external hard drive on site for a week, getting a week's worth of daily backups, then replace that hard drive with another one and take the first drive off-site.
The next backup is more of an administrative and overall comprehensive backup. For this I usually use Windows Server Backup on the physical host. Attach an external hard drive, network share, or even just a partition on the same hard drive as your above daily backups. And set up Windows Server Backup to automatically backup your server daily to that space. It should get the host OS, the virtual machines, your critical data, everything. So in the event I need to restore my server on the same hardware and don't want to transfer a ton of data individually, I can just to a system recovery using the Windows Server 2012 DVD to perform a backup recovery. I have had this fail on two rare occasions though, so I don't always trust it 100% with all the data. After all, in an absolutely disastrous failure I can always go in and re-build the server, rebuild the domain controllers and other virtual machines, but it is the business critical data that cannot be rebuilt near as easy.
High availability can be a huge benefit for any business who wants to minimize downtime as much as possible. If you look at your company and start crunching the numbers and listing out the services and people who are affected if your servers go down, it can be quite shocking to see how critical of a role they play in daily business life anymore. This is why businesses with 50 or more employees who are looking at new servers may be considering HA. With this situation, you would set up a SAN device of some sort (either a server, SAN storage device, or scale-out file server cluster) which would actually hold the data for your virtual machines. This SAN network is accessible from multiple host servers but not always simultaneously. So, if you have Server1 running your SQL database and suddenly Server1 has a power supply fail and shuts off, the cluster would detect this failure and start up the virtual machine for your SQL database again on your Server2 system meaning very short downtime and little to no data loss.
That being said, HA is definitely not something to jump into just because of its benefits because it's caveats can be even more critical. For example, what happens if your SAN server suffers a power supply failure, motherboard failure, RAID controller failure, corrupted partition, etc. etc.? The heart of your system is really that SAN and if it goes down you have suffered a cataclysmic failure, also known as the inverted pyramid of doom. SANs can be unbelievably expensive. There are many types out there, but you're looking at a minimum of $10k - $15k just to get a basic SAN system going. That leaves you no budget for the necessary network infrastructure, server hosts, or licensing. SAN and cluster environments are also more complex to manage, configure, and maintain overall.
Replication is the cheap man's version of HA. It's making duplicates of the data and changes to that data at regular intervals to another server. In the event that the primary server goes down, the virtual machine has to be manually started up on the other server system, so it's not automatic. There can also be several minutes worth of lost time or data, as it will go off the last good replicated datapoint when the server starts up again. However, setting up Replication only has cost in the actual host servers, there is no need for extra SAN storage and configuration.
One thing I'm not completely positive on, but from what I understand is replicated VMs do not require an additional license and will work off the original VM's Windows Server 2012 license so long as you aren't running both simultaneously of course. So in the situation that you are talking about above, you'd have Server1 with a Windows Server 2012 license, and Server2 with a Windows Server 2012 license, and this would actually give you a total of four VMs, two on each server. That means you should be able to set up Server1 with SQL server, a backup domain controller, and Server2 with your primary domain controller and perhaps an application server or something else as needed.
When it comes to replication, it can be pretty forgiving about varied hardware. For instance, for a live migration the servers have to have the same type of processors (Intel or AMD) and it is recommended that they have the same features or are the same platform (for instance, LGA2011.) However, they don't have to be identical. From the information and research that I have read on it, however, it is recommended to keep the servers as close to identical as possible (RAID controllers, NICs, processors, etc.) This eliminates compatibilities moving the VMs between different underlying hardware that may require different driver support, etc. I also recommend keeping them the same processor platform (socket type, chipset, etc.) because of physical hardware support. If you need to buy additional RAM the same RAM is compatible in both servers that way. If you need to upgrade processors they are the same compatible processors.
When it comes to backup again there are numerous ways and products to do so. I've tried out the free version of VEEAM and it worked pretty well but didn't have any of the automatic features I was looking for without purchasing a full version of the software which none of my customers were for given the cost. Mostly what I use is a two-fold backup system. First, within the VM. For example with a file server virtual machine, I utilize a data copy application like Uranium Backup (which is free) to schedule automatic backup of all the actual business data (documents, pictures, etc.) to a network shared location or additional external hard drive. There are a TON of options here within Uranium, such as doing daily backups for a full week, then overwriting the backups day by day the next week. This gives you a full week's worth of previous versions to go back to if necessary but obviously increases the amount of storage capacity you need.
This is my critical data backup. With this backup I have all the original files in their original individual formats and not in some proprietary compressed file. So, in the event that everything went haywire and I had to get access to some files within minutes for someone, I can unplug the external hard drive, plug it into any other computers, and I have access to all of the company data. Ideally this should be set up in a rotation. For instance, have one external hard drive on site for a week, getting a week's worth of daily backups, then replace that hard drive with another one and take the first drive off-site.
The next backup is more of an administrative and overall comprehensive backup. For this I usually use Windows Server Backup on the physical host. Attach an external hard drive, network share, or even just a partition on the same hard drive as your above daily backups. And set up Windows Server Backup to automatically backup your server daily to that space. It should get the host OS, the virtual machines, your critical data, everything. So in the event I need to restore my server on the same hardware and don't want to transfer a ton of data individually, I can just to a system recovery using the Windows Server 2012 DVD to perform a backup recovery. I have had this fail on two rare occasions though, so I don't always trust it 100% with all the data. After all, in an absolutely disastrous failure I can always go in and re-build the server, rebuild the domain controllers and other virtual machines, but it is the business critical data that cannot be rebuilt near as easy.
-
Reply to choucove
m
0
l
ropern01
July 14, 2014 10:33:12 AM
10-4 on that. Great info about everything , especially on the backup method. Can anyone verify what Choucove is saying about the license with live migration/replication when the machines don't run simultaneously?
Choucove, your last post has got me thinking about my decision to purchase the different type of servers. I think I am going to try and get 2 R420's. May just have a tad more memory and maybe that better array controller in the server that runs the SQL. I don't want to skimp on anything because I never want to have to have the failover situation. However in order to get alike servers in under budget I may have to pass on some of the insurance features like say redundant power supplies. After all there is a SLA there in case a part goes out and with an alike server being in house there is a good chance the vm will come up without many issues. I will definitely be doing a lot of testing this before implementation. I have messed around enough with VMware to know what can happen sometimes with the different hardware. However just recently I was able to convert an Acronis backup into a virtual hard drive and get the machine up and running. Truly was a life saver in that particular situation. OK, good stuff.
One last topic I haven't mentioned but I have been thinking about is now that we are going virtual with this client I am considering a maybe a three year lease option. I feel like the servers have become a lot more mobile now that they are not stuck with having to be installed on alike hardware. I guess I should get out a calculator once I get some figures and see what the total cost of ownership will be with a lease. Also that way I never have to worry about what to do when we have a hardware failure. We will always have a SLA. I wonder if that covers hard drives? Which gets me thinking of one more issue I saw you talking about in another post (Choucove). If I am close to budget after skipping some of the redundancy items, I may consider purchasing my own drives to save some dinero. I wonder how that affects the warranty of these servers. Guess I could ask my sales team but I am never really all trusting when it comes to sales teams. Some of you may be wondering why doesn't this guy just ask the vendor. Well In my opinion that is one of the main issue when purchasing one of these machines . You are never truly sure about the answers you get because it's their job to try and sell you this great equipment. While they may be selling you some great equipment you may not necessarily need it and that equates to big bucks.
OK I am getting very close to pulling the trigger on the server and infrastructure purchase. I am ready to get this project rolling. Seems like I have been researching this stuff forever. I wanna give a HUGE thanks to you Choucove for assiting me with this project. Your insight truly has helped me get over the hump on deciding which way to go with some of these delicate decisions. You are the man!!!!
Once I go through this process I will be able to replicate it a lot easier. Maybe when I get everything up and running and I am able to reflect on my decisions I can put a little list of exact hardware and software used with time and prices. That can be my contribution for anyone else who wants to do this kind of implementation but is having trouble putting it all together.
Choucove, your last post has got me thinking about my decision to purchase the different type of servers. I think I am going to try and get 2 R420's. May just have a tad more memory and maybe that better array controller in the server that runs the SQL. I don't want to skimp on anything because I never want to have to have the failover situation. However in order to get alike servers in under budget I may have to pass on some of the insurance features like say redundant power supplies. After all there is a SLA there in case a part goes out and with an alike server being in house there is a good chance the vm will come up without many issues. I will definitely be doing a lot of testing this before implementation. I have messed around enough with VMware to know what can happen sometimes with the different hardware. However just recently I was able to convert an Acronis backup into a virtual hard drive and get the machine up and running. Truly was a life saver in that particular situation. OK, good stuff.
One last topic I haven't mentioned but I have been thinking about is now that we are going virtual with this client I am considering a maybe a three year lease option. I feel like the servers have become a lot more mobile now that they are not stuck with having to be installed on alike hardware. I guess I should get out a calculator once I get some figures and see what the total cost of ownership will be with a lease. Also that way I never have to worry about what to do when we have a hardware failure. We will always have a SLA. I wonder if that covers hard drives? Which gets me thinking of one more issue I saw you talking about in another post (Choucove). If I am close to budget after skipping some of the redundancy items, I may consider purchasing my own drives to save some dinero. I wonder how that affects the warranty of these servers. Guess I could ask my sales team but I am never really all trusting when it comes to sales teams. Some of you may be wondering why doesn't this guy just ask the vendor. Well In my opinion that is one of the main issue when purchasing one of these machines . You are never truly sure about the answers you get because it's their job to try and sell you this great equipment. While they may be selling you some great equipment you may not necessarily need it and that equates to big bucks.
OK I am getting very close to pulling the trigger on the server and infrastructure purchase. I am ready to get this project rolling. Seems like I have been researching this stuff forever. I wanna give a HUGE thanks to you Choucove for assiting me with this project. Your insight truly has helped me get over the hump on deciding which way to go with some of these delicate decisions. You are the man!!!!
Once I go through this process I will be able to replicate it a lot easier. Maybe when I get everything up and running and I am able to reflect on my decisions I can put a little list of exact hardware and software used with time and prices. That can be my contribution for anyone else who wants to do this kind of implementation but is having trouble putting it all together.
-
Reply to ropern01
m
0
l
Going with third party hard drives can possibly save you noticeable costs when going with a server system like this, especially if you have multiple hard drives that need to be ordered or even if you are considering SSDs. After all, the hard drives used by HP/Dell, etc. are just Seagate, Western Digital, or Hitachi enterprise hard drives with a slightly customized firmware image to make them appear as the OEM drive from HP or Dell. However, those drives come at about a 200% price premium compared to the original drive's price.
That being said, it does come with a slight risk involved. Hard drive warranties, especially for enterprise hard drives, are better going directly through the OEM, but I'd still recommend having one spare on hand so you have immediate replacements if needed. But if you are looking at complete stability and compatibility, it's true that the HP hard drives have been tested and ensured through guarantee to work with the HP servers, RAID controllers, and specific drivers/firmware as well. Basically if you buy an HP hard drive and have it in an HP server and something buggy comes up with the drive firmware while operating a certain array type on your specific RAID controller, it's up to HP to fix it. But, if it's not specifically the HP drive, they may say, "Well it's not a verified drive." Likewise, if there is a support issue and perhaps you have an issue with a random drive appearing and disappearing from the array but the drive tests fine, HP tech supports COULD say, "Well you don't have an HP hard drive so you need to fix that as a possible compatibility issue." Now, I've never come across that yet, but perhaps I'm just lucky. In the end my main suggestion is either way you go, ensure that you are getting a good quality enterprise hard drive and not a basic desktop or low power drive. If one of those cheaper drives fails quickly all you will probably get from the vendors is that those drives weren't meant for servers anyways so it's your own fault.
That being said, it does come with a slight risk involved. Hard drive warranties, especially for enterprise hard drives, are better going directly through the OEM, but I'd still recommend having one spare on hand so you have immediate replacements if needed. But if you are looking at complete stability and compatibility, it's true that the HP hard drives have been tested and ensured through guarantee to work with the HP servers, RAID controllers, and specific drivers/firmware as well. Basically if you buy an HP hard drive and have it in an HP server and something buggy comes up with the drive firmware while operating a certain array type on your specific RAID controller, it's up to HP to fix it. But, if it's not specifically the HP drive, they may say, "Well it's not a verified drive." Likewise, if there is a support issue and perhaps you have an issue with a random drive appearing and disappearing from the array but the drive tests fine, HP tech supports COULD say, "Well you don't have an HP hard drive so you need to fix that as a possible compatibility issue." Now, I've never come across that yet, but perhaps I'm just lucky. In the end my main suggestion is either way you go, ensure that you are getting a good quality enterprise hard drive and not a basic desktop or low power drive. If one of those cheaper drives fails quickly all you will probably get from the vendors is that those drives weren't meant for servers anyways so it's your own fault.
-
Reply to choucove
m
0
l
ropern01
July 20, 2014 7:30:32 PM
Sorry I disappeared there for a minute. Got smacked around a little by life...lol. In a matter of a week my vehicle was down, phone went down, refrigerator went down, John Deere went down and my dryer went down. I am scared to order the servers now!!! LOL. Anyway I couldn't agree with you more Choucove. Anyone who has been in this business long enough has been bounced around by some vendor/vendors who point the finger or place the blame anywhere else if given an opportunity. I want to have at least one machine that I could count on them fixing anything wrong hardware wise with no excuses.
So here is the plan. It is starting to look like buying the alike machines is going to put me over budget unless I drop down a model or skimp on something. So, I think I may just get one direct from Dell and the other from one of the more reputable server dealers on eBay. I saw the same model with almost three full years warranty from Dell for a lot cheaper. I could put my own drives in it saving even more money and roll the dice with warranty repairs on this box. That should leave me enough room in the budget to buy everything from them on the most powerful server. And if anything goes out on the cheaper box, the more powerful machine hopefully run everything without much lag. If the main server goes down I have the other Dell which should be able to get us by reasonably well. This method should keep me relatively close to my goal in terms of budget as well as the peace of mind that if something happens I have full coverage on the direct purchase box for sure. And Hopefully somewhere down the road we can find a steal on a third machine for insurance. You can never have to many backups or backup plans when these things go wrong. I know I keep saying it but I really appreciate all the help. I was really struggling with the many delicate decisions and you really helped.
PS. I am wondering if you have a pay-pal acct or something where I could send you a little consultant's fee when I finish this job and get paid. If so you can pm or you have my email address, just let me know which method you would prefer if any. If not I understand because you don't know me from Adam. I wouldn't be offended if you just rather not but I just wanted you to know how much it meant. I would offer it now but I am just struggling mightily with everything that's happened lately. It's just refreshing to see that people like you are actually still out there. It leaves some hope for humanity!!!
Thanks again!!!
So here is the plan. It is starting to look like buying the alike machines is going to put me over budget unless I drop down a model or skimp on something. So, I think I may just get one direct from Dell and the other from one of the more reputable server dealers on eBay. I saw the same model with almost three full years warranty from Dell for a lot cheaper. I could put my own drives in it saving even more money and roll the dice with warranty repairs on this box. That should leave me enough room in the budget to buy everything from them on the most powerful server. And if anything goes out on the cheaper box, the more powerful machine hopefully run everything without much lag. If the main server goes down I have the other Dell which should be able to get us by reasonably well. This method should keep me relatively close to my goal in terms of budget as well as the peace of mind that if something happens I have full coverage on the direct purchase box for sure. And Hopefully somewhere down the road we can find a steal on a third machine for insurance. You can never have to many backups or backup plans when these things go wrong. I know I keep saying it but I really appreciate all the help. I was really struggling with the many delicate decisions and you really helped.
PS. I am wondering if you have a pay-pal acct or something where I could send you a little consultant's fee when I finish this job and get paid. If so you can pm or you have my email address, just let me know which method you would prefer if any. If not I understand because you don't know me from Adam. I wouldn't be offended if you just rather not but I just wanted you to know how much it meant. I would offer it now but I am just struggling mightily with everything that's happened lately. It's just refreshing to see that people like you are actually still out there. It leaves some hope for humanity!!!
Thanks again!!!
-
Reply to ropern01
m
0
l
Related resources
- Small Business Server - Need Help picking Forum
- Small Business Server, Help Forum
- Using small business server as a desktop ( problem! help! ) Forum
- Small business server help Forum
- Need small business server hardware help Forum
- Small Business Server build help Forum
- Small Business Server Build - Help Request! Forum
- small business network + file server ... HELP! Forum
- Small Business Server Build Help! Forum
- More resources
Read discussions in other Business Computing categories
!
Guess I should have went with the one liner!!!