Through a unique set of circumstances we were recently lucky enough to end up with two Intel 910s in our lab. We certainly could not pass up the opportunity to put both of these PCIe SSDs together to explore the limits of performance that can be reached with two of these powerful drives in tandem.
With one of the key selling points of the Intel 910 being its price, many customers will use more than one of these high performance SSDs in a single server to solve performance problems or to accelerate existing systems.
We did cover the Intel 910 more in-depth in our initial article, the Intel 910 800GB and 400GB PCI Express Solid State Drive Review. This article will provide a primer for those who are unfamiliar with the Intel 910.
In this article, we will merely be getting right down to benchmarking two of these impressive PCIe SSDs.
The Intel 910 aims to be a fairly easy device to configure and setup for end-users, with a very low overhead driver stack and easy to configure volumes. This certainly holds true, even for a multiple SSD setup. Once again it was just as simple as plug and play to get the hardware up and running with two of the SSDs. The only task left was to configure the volumes in the operating systems Disk Management utility.
In today's article, we will take a quick look at the speed of two of these devices together in an enterprise scenario. One implementation that the Intel 910 seems to be uniquely well suited to is in the very demanding OLTP (On-Line Transaction Processing) field. These types of workloads feature drive activity under some of the most punishing scenarios, with a heavy read/write percentage. The OLTP sector typically prizes performance over all other factors, including cost. The 910 looks to change this by delivering great endurance and high performance in a much more affordable package that would lend itself well to multiple card applications.
We will be performing simulations of this type of scenario and looking at some of the other types of applications that these SSDs may be deployed into.
Specifications, Pricing and Test System
Specifications, Pricing and Availability[img]3[/img]
Each drive is rated for 180,000 random read and 75,000 random write IOPS, along with 2GB/s sequential read and 1GB/s sequential write for the 800GB model. The 910 specifications are measured when the SSD is under steady state conditions and filled to 100% capacity. Intel does not market F.O.B. (Fresh Out of Box) specifications for the 910.
There is also the option to select a "Performance Mode' for this device. This allows the user to select a higher average power draw for the SSD which results in a sequential write performance increase up to 1.5GB/s from the standard 1GB/s. This does increase the average power draw from 25 Watts up to 28 Watts and the peak output to 38 Watts. Our testing with both SSDs will be at the default power settings.
The 800GB model has an MSRP of $3859 and the 400GB is listed for $1929. This is an impressive price point, coming in at $4.82 per GB, outstripping much of the competition easily. This low price point is one of the primary reasons that we see the Intel 910 being very successful in the datacenter space.
Test System and Methodology
The Intel 910 comes in both 400GB and 800GB capacities. The 800GB 910 presents itself to the host operating system as four individual 200GB volumes, while the 400GB version only supplies two of the 200GB volumes. We have tested both in a previous article and will be providing those results as a baseline comparison to the performance of the dual 910s.
There is the possibility of using Windows RAID or a third party program, to aggregate the performance of all of the drives into one large RAID 0 volume. This would be a risky deployment in most applications, but when paired with strong backup schemes in high read and write scenarios, RAID 0 can be a compelling solution. Parity can also be achieved with RAID 5, providing data redundancy at the sacrifice of capacity and write speed.
The configuration that we have tested with for the enterprise portion provides the best latency results and does a good job of showing the base performance of the dual 910's. This is simply configuring it as eight separate LUNs and accessing each individually. This provides low overall latency in conjunction with much lower maximum latencies than RAID 0. This portion of testing also adheres to our testing regimen that is based upon the SNIA specification. More information on our enterprise testing regimen can be found here.
Base Product Specifications
The five major measurements of base performance of any Solid State Storage solution are latency, random read/write and sequential read/write speed. These are the most common measurements that are posted by manufacturers to advertise storage performance. We will cover the latency and sequential read and write on this page and random performance on the following page. We begin with a measurement of the latency of the device. The industry standard for measurement of latency is 4K Random Access at a Queue Depth (QD) of one.
The Dual 910s access time is virtually identical to the 800GB single device measurements, causing them to actually be hidden underneath the 800GB results. The Server 2008 R2 LSI drivers handle all 8 volumes seamlessly even with two PCIe SSDs connected, so there is virtually no impact to base latency for either device.
The 128K sequential read speeds reflect the maximum sequential throughput of the SSD using a realistic file size that will actually be encountered in an enterprise scenario. With two 910s aggregating their speeds they manage to push 3,928MB/s, slightly under 4GB/s.
The 800GB scores at 1956MB/s, barely below the advertised 2.0GB/s. We can observe similar results with the 400GB, coming in at 973MB/s, a tad below the rated 1GB/s. Once again, as with the 400 and 800GB single SSD results, we will observe nearly perfect scaling over the course of the benchmarks. The LSISAS2008 PCIe bridge scales exceptionally well with each of the four controllers on each SSD, so the dual 910 results are at expected levels in almost all benchmarks.
The two 910s reach 1820MB/s in sequential write speed. This is not perfect scaling with the 800GB results, which is due to running the dual tests at the default power setting. The 800GB tests were conducted at the maximum performance setting. These results are roughly 180MB/s below the expected value for the two SSDs together in the default performance mode.
The 128K Sequential Write speeds of the 800GB, in maximum performance mode, top out at 1480MB/s, right below the rated 1.5GB/s using the Maximum Performance mode. The default power mode hits 1GB/s with sequential write access. The 400GB reaches 745MB/s, right at the rated spec of .75GB/s.
4K and 8K Random
4K random read speed measurements are an important metric when comparing drive performance, as the hardest type of file access for any storage solution is small-file random. We test with a five-second ramp time to eliminate any burst results. Read speeds are largely unaffected by extra over-provisioning, so we only include standard results.
The 4K random read IOPS scaling is perfect, with actually a slightly better than linear scaling with the two 910s topping out at 457,146 IOPS. This represents pulling off 228,573 IOPS per SSD, well above the Intel-advertised 180,000.
The 800GB tops out at QD256 at 227,247 IOPS, this is also above the rated 180,000 IOPS by over 21%.
The 400GB version peaks at 113,489 IOPS at QD128, also 21% over the rated specification of 90,000 for that capacity.
As we move into the random write results, we begin to include our 20% over-provisioning results, indicated as "OP" in our graphs. OP consists of leaving 20% of the drive unformatted. By increasing the amount of available spare area, the SSD can perform at higher write speeds. This represents an optional performance boost for users at the cost of some capacity.
The 910s do not disappoint, with a top speed in steady state of 164,946 IOPS at QD32. The over-provisioned (OP) results top out at 186,839 IOPS at QD32. This scaling does represent a slight loss in IOPS attained per device.
The 800GB results with standard Steady state peak at 85,768 IOPS at QD32. This is 13% higher than the rated 75,000. The OP result reaches 100,342 IOPS at QD32. The 400GB plateaus at QD32 with 42,971 IOPS, 12% higher than the rated 38,000. The OP results reach 52,421 IOPS at QD16.
8K random read and write speed is a metric that is not commonly tested for consumer use, but for enterprise environments this is an important aspect of performance. With several different workloads relying heavily upon 8K performance we include this as a standard with each evaluation. Many of our Server Emulations below will also test 8K performance with various mixed read/write workloads.
For 8K 100% Random Read the dual 910s blast out 213,426 IOPS, again actually coming out with better than linear scaling.
The 800GB version scores 104,700 IOPS at QD128 and the 400GB version tops out at 52,166 IOPS at QD128.
The 8K Random write speed in Steady state with the dual 910s reaches 44,360 IOPS and the extra 20% of Over-provisioning boosts that value up to 49,146 IOPS. There is a slight loss with the 8K random writes in IOPS attained per SSD. The 800GB model tops out at QD64 with 23,995 IOPS and gets a boost up to 26,000 with extra Over-provisioning. The 400GB model scores 12,275 at QD32 in Steady state and with extra Over-provisioning scores 13,607 at QD32.
This test emulates Database and On-Line Transaction Processing (OLTP) workloads. OLTP is in essence the processing of transactions such as credit cards and is used heavily in the financial sector. Enterprise SSDs are uniquely well suited for the financial sector with low latency and high random workload performance. Databases are the bread and butter of many enterprise deployments and their data access patterns are similar to OLTP. These are demanding workloads with 8K random of 66% read and 33% write distribution that can bring even the highest performing solutions down to earth.
The dual 910s reach 668MB/s in Steady state at a QD of 128 and with extra OP that value is raised to 719MB/s. The 800GB model reaches 344MB/s at QD64 with regular Steady state and 359MB/s with OP. The 400GB model reaches 172MB/s in Steady state at QD64 and 178MB/s with OP.
This test emulates a typical Email Server with a 50% Read and 50% write distribution of 8K random files. These types of servers aren't typical applications that many would expect solid state storage to be integrated into. With the lowering price points of high endurance SSDs, acceleration for these types of servers is becoming a reality. An added bonus is that an enhanced amount of IOPS capability equates to a higher average CPU utilization of the server, maximizing the performance and TCO/ROI of the overall system.
The dual 910s provide 560MB/s in Steady state at QD128 and 620MB/s with Over-provisioning. The 800GB model scores 290MB/s at QD64 with normal Steady state and 305MB/s with extra OP. The 400GB model reaches 221MB/s at QD32 in Steady state and 155MB/s at QD32 with OP.
The File Server profile represents typical workloads that will be encountered in file servers. The inherently random nature of data access in file servers can gain considerable boosts in performance from SSDs and also benefit greatly from caching and tiering solutions. This profile tests across a wide variety of different file sizes simultaneously, with an 80% read and 20% write distribution.
Again we can observe great speeds with the dual 910s achieving 876MB/s in Steady state and 932MB/s with OP.
The Web server profile is a read-only test with a wide range of file sizes. Web servers are responsible for generating content for users to view over the internet, much like the very page you are reading. The speed of the underlying storage system has a massive impact on the speed and responsiveness of the server that is hosting the websites and thus the end user experience.
The dual 910s provide 1945MB/s with both Steady state and OP at QD128. The 800GB model scores 976MB/s at QD64 in Steady state and 995MB/s with OP. The 400GB model scores a similar 970MB/s in Steady state and 993MB/s with OP.
TCO (Total Cost of Ownership) is the big story here. Price has always been one of the major inhibitors for companies looking to upgrade to enterprise-class flash products and Intel is helping to break down those boundaries. One of the great benefits of more value-centric solutions such as the 910 is the ability to use them in applications where many users are not commonly accustomed to seeing flash devices employed.
OLTP, VMware, Cloud Computing and other high impact I/O applications are the most common implementations for PCIe SSDs and rightly so. The latency and tremendous random access speeds that solid state storage can deliver fit right into the needs of these I/O intensive applications. In many of these applications, particularly the financial sector, cost is no object and performance trumps all other considerations.
As solid-state storage becomes more affordable, there will be greater penetration into servers that were previously not considered for this type of high performance flash device. Caching and Tiering models in particular can easily accelerate existing infrastructure. Using the Intel 910 as a front end for a large array of HDDs can provide instant results. Through a variety of software caching programs, some of which we will be testing with the 910 in the coming weeks, hot files can be placed onto the SSD from the HDDs providing massive improvements in existing servers. This in turn keeps the CPU utilization in the host server high, maximizing the performance of the entire system.
A growing concern is also the increasing network traffic and infrastructure required in the datacenter. As many more robust caching software programs make their way into the datacenter look for this traffic to be cached and accelerated using flash storage as well.
With so many uses for flash in the server space, we look to see devices such as the Intel 910 deployed en masse in the very near future and the testing today gives us a taste of the performance that can be expected from multiple PCIe SSDs. The small form factor, ease of use and lower price points will surely make applications such as this much more commonplace in the future.
With the marriage of Intel's own flash and controllers to the LSISAS2008 PCIe Bridge and drivers, Intel has definitely delivered in the performance aspect, beating all of their own marketed specifications easily. We easily attained 457,146 4K Random read IOPS. The 4k Random write IOPS reached an impressive 178,482 in Steady state. The capability of 191,102 4K random write IOPS in Steady state with 20% over-provisioning at QD128 tells the story for the possibility of configuring the device for high random workloads as well.
The Intel 920 is truly a well-rounded performer, in both single and multiple applications.
PRICING: You can find products similar to this one for sale below.
United States: Find other tech and computer products like this over at Amazon.com
United Kingdom: Find other tech and computer products like this over at Amazon.co.uk
Australia: Find other tech and computer products like this over at Amazon.com.au
Canada: Find other tech and computer products like this over at Amazon.ca
Deutschland: Finde andere Technik- und Computerprodukte wie dieses auf Amazon.de