- Adaptec's Series 5 Unified Serial Controllers
- iSCSI SANs Compared
- Can Heterogeneous RAID Arrays Work?
- Storage Accessories: Move, Cool, Multiply
- Portable Storage: Convenience is the Key
- AMCC, Areca & LSI Serial RAID Controllers
- RAID Scaling Charts, Part 3: 4-128 kB Stripes Compared
- Take Your Storage Online
- Is Zettabits Next Generation Storage?
- Unified Serial RAID Controllers for PCIe
Conclusion
Source: Tom's Hardware – Keywords: ssd, memoright
Syndication:
Conclusion
The benchmark results for the Memoright flash SSDs (we used four 32-GB models MR25.2-032S) speak for themselves. A sequential throughput of 115 MB/s is a new record for flash-based drives, and Memoright even managed to sustain almost the same throughput for write operations as well. I/O performance is stellar and the drive’s power consumption is lower than the power requirements of the direct competitor, the Mtron Flash SSD. The only benchmark sections where it cannot beat everything else is the PCMark05 Windows XP startup benchmark and the IOmeter Webserver benchmark. In every other test, Memoright slaps the other drive manufacturers in the face by providing bone-crushing storage performance. Server administrators should especially study the benchmark results carefully, as we’re talking about many hundreds to thousands of I/O operations per second on an individual drive.
Memoright’s Flash SSDs are also very RAID-friendly, as they worked properly with an Adaptec RAID 5805. We decided to use four Memoright flash SSDs to compare against four Seagate Savvio 10K.2 drives and four Western Digital Raptor WD1500 drives – all in RAID 0, so we could determine the absolute maximum performance. The result didn’t come as a surprise, but it still was an eye opener: The Memoright drives are significantly superior providing multiple times better I/O performance and up to 50% more throughput for streaming applications.
Clearly, this Flash SSD is the best system hard drive you can possible use. But it has one major disadvantage, which is its price point. A capacity of 32 GB costs as much as $1,000 and the higher capacity versions, which would be very nice to have, are even more expensive. Although the 128-GB model provides the best cost per gigabyte ratio, all these drives are way too expensive. Most users would not even spend $1,000 on their entire PC. If you can live with the fact that these drives will probably cost half of today’s cost or maybe only a third by the end of this year, then you can go for it, as you can be sure to get the very best hard drive available.
The transition from conventional HDDs to Flash SSDs has begun, as these results are convincing enough to make decision makers rethink their storage media strategies. If fast throughput rates are especially critical for a particular enterprise’s datacenter, for example, then the high costs required to invest in high-performance SSDs can make sense. It’s clear that 15,000 RPM hard drives will be among the first victims, as I’d rather go for a small-and-fast flash SSD plus a large storage drive instead of a single, medium capacity 15k SAS drive for a server.
But all of this will remain in the high-end or enterprise segment for the time being, as it won’t be possible to replace all 500 million hard drives sold in 2007, or even only all server and system drive with flash-based hard drives for another very simple reason: According to Western Digital, all the flash production sites in the world aren’t sufficient to produce enough memory to replace that of traditional hard drives. Hence, we will have to wait for further improvements in flash memory density and affordable 100+ GB flash drives to drop below $500 before they enter the mainstream.
- Previous page I/O Performance (RAID 0, 4 Drives)
- Next page Flash SSD Comparison Table
- 1 / 2
- Next
-
The key here is that SSDs are optimum for big retailers & enthusiasts, because of the lower failure rate compared to mechanical drives means less hardware failure and RMA and tech support calls. For us, its performance based. Dell and others will flock to these when the pricepoint hits home, and as soon as they jump on, drives will sell in such volume that new plants will have to be made and we'll need SATA-1000 before long.
Anyone whose been in the industry professionally and watched tech roll out over the years can see the writing on the wall, SSDs will dominate at least the "bootdrive" sector within two years. It might take longer to ramp up capacity for the "data" drives for home users, but its coming, and every process shrink to a smaller fab makes capacity go up exponentially.
The ONLY, and I mean only, area where ssd drives will get consumer exposure will be in the laptop sector, simply because of the size/power requirements. For that to happen, ssd prices will need to fall "tenfold" and conventional hard drives will need to die out, which won't happen any time soon.
Desktops will will continue to use conventional hard disks for at least the next 5-10 years as they will go down in price and increase in capacities. Sata III 6gbps and newer pmr technologies will keep the traditional hard drive alive for a long time yet.
Servers may transition into a hybrid environment with both ssd and conventional drives, but with the cost of raid now so affordable (controllers incorporated into chipsets) and the fact that traditional fast (velociraptor) and beefy (terabyte) drives will ALWAYS be cheaper than the ssd variants - it's going to be a hard sell.
I'm not saying ssd is bad, but for those who don't have 5-digit monthly income, it is merely a toy for the rich. Sure, prices will eventually fall, but until you can sell the technology for 25 cents/gig or better, it will remain insignificant.
| geralt : SSDs have weakness in random writes and reads. Unfortunately this was not tested here. |
A mix of SSDs and HDDs will probably be the best solution for a long time. They each have their strengths depending on usage patterns. As soon as SSDs become more economical, mixed setups will quickly become the norm. I would have a couple of SSDs in a RAID 0 config to boot on and load applications off of, but I would have a some HDDs in a RAID setup for mass storage.
I really can't wait to build my next system. Gone is the day where I reuse a harddrive on a new build, and the new computer seems only as fast as the old one due to the HDD bottleneck.
As someone else commented on I am HORRIBLY disappointed that Tom's didn't test random writes and reads. If they actually didn't that is, I haven't finished the article but wanted to respond to the "clown" guy. I've noticed a lot of stuff like that with tech sites and it makes me wonder what kind of enticements these guys have to talk up a given product. Please do a follow up with random read&write tests.
as fas as the size goes. there is only a small amount of XP/Vista that needs benefits from super fast access. Your application and it's data are usually the best canidates for super fast storage.
/And yes show us the random access tests since you do talk about OS storage.
Give me a break. Maybe years from now but that's not exactly news. Anyway, when it finally happens this specific model will not be the one to do it. It may end up being an entirely different technology that ends the HDD if SSD's don't drop in price fast enough.
Drop the sensationalism, start labeling both the X and Y axis of your graphs like professionals, and try to regain some journalistic integrity.
Toms is going downhill fast.
In fact, even if the 32gb ssd drive was $300, I would still have a hard time justifying the cost for such little space. 128GB at $300-$500 and I MAY think about buying one or more of these.
- 1 / 2
- Next
-

This has about the same marketing hype as ddr3 compared to ddr2.
I could care less about access times.
I am not paying their prices for a 32gig "SSD".
It is just not worth the money.