Here’s Seagate’s new lineup:

• Seagate 600 6Gb/s SATA, a drive that Seagate calls: “The ultimate laptop upgrade”.  The company claims that this is the first 5mm-high 2.5″ SSD on the market (although it is also offered in a 7mm thickness).  It comes in 120, 240, and 480GB capacities and can support 40GB of writes per day.

• Seagate 600 Pro is also a 6Gb/s SATA SSD, aimed at lowering power consumption in servers and storage by delivering 10K IOPS per Watt when presented with a 50/50 read/write balance and adjusting its power consumption to the workload.  Seagate calls this an “entry-level enterprise SATA SSD offering best in class IOPS/Watt.”  The product has been tuned to provide consistently low latency under balanced workloads, and has a tight latency distribution when presented with a 70/30 read/write mix.  This SSD ships in capacities of 120, 240, and 480 GB for lower write-load applications, and in 100, 200, and 400GB capacities that support 2 drive writes per day.  To appeal to the data center this drive offers a 1.8 million hour MTBF and offers error recovery and power-loss data correction.

• Seagate 1200 is a 12Gb/s SAS SSD.  This device has been designed for write-intensive mixed workloads, and uses sophisticated algorithms to optimize data placement and prioritize reads and writes.  The self-encrypted SSD is offered through distribution channels in a 2.5″ form factor and to OEMs in both 1.8″ and 2.5″.  Its 12Gb/s SAS interface is backwards-compatible with standard 6Gb/s SAS ports.  The 1200 comes in two models – the base model (supporting 10 drive writes per day) in capacities of 200, 400, and 800GB, and the high endurance model (offering 25 drive writes per day) in capacities of 100, 200, and 400GB.  Like the Seagate 600 Pro above, this SSD tunes its power consumption to the workload.  Seagate is seeking FIPS certification for this model.  The company is proud of the fact that the controller was designed 100% internally by Seagate to use Samsung’s 21nm flash leveraging the Seagate/Samsung strategic agreement.

• Seagate X8 Accelerator is Seagate’s brand for the Virident FlashMAX II PCIe SSD, but one that has been put through the full suite of Seagate’s quality assurance tests.  This 1.1 million IOPS SSD is offered in 550GB, 1.1TB and 2.2TB capacities in a half-height, half-length (HHHL) PCIe form factor and offers an average latency below 20 microseconds.  End-to-end data protection supports data integrity in this 8-lane device.  One interesting aspect of this PCIe SSD is that it is scalable – multiple drives can be added to a server to appear as a single very large volume.

All of the new SSDs are engineered to deliver ultra-fast speed and high data integrity, to protect against data loss in power failures or other corruption.

Objective Analysis publishes a number of market reports on Client and Enterprise SSDs that can be purchased for immediate download on the web.  For a full listing of these reports visit the Objective Analysis Reports page.

The report is based upon a survey that asked IT managers about their enterprise IOPS requirements.  The webinar gives a taste of the report’s contents, and explains the survey methodology.  During the course of the webinar and at the end Tom and I answered a number of listener questions relating to the content.

The presentation also includes a little plug for SNIA’s client IOPS survey which is being run by downloading a program called the Workload I/O Capture Program, or “WIOCP.”

A replay of this webinar is available on the BrightTalk website.

The presentation was well received by our audience.  Have a listen.

In a nutshell, co-founders David Flynn (CEO) and Rick White (Chief Marketing Officer) have left the firm, with Flynn being replaced by board member Shane Robison, HP’s former Chief Strategy Officer, who helped put together HP’s acquisition of Autonomy.

Here are some of the points made in the Alert:

• Flynn and White will remain on the board for another 12 months
• Robison told Bloomberg that there was “no disagreement on strategy” behind the change
• Fusion-io has given some advance notice for prior friendly CEO transitions, which was not the case this time
• In its release Fusion-io took pains to make clear that this wasn’t the same kind of leadership change that recently occurred at OCZ and sTec
• The company’s strategy has kept it a step or two ahead of its competition.  Its success will hinge on continuing in this direction.

In our report: The Enterprise SSD: Technologies and Markets, Objective Analysis forecast that the PCIe SSD market would consistently out-perform the SAS SSD market.  The SSD Guy certainly hopes that Fusion-io can continue to retain its prominent position in this promising market despite the change in leadership.

The SSDs are MLC products, rather than the SLC ones used in earlier K2s, allowing Kaminario to reduce the cost.  Although the SSDs Kaminario uses come with a 5-year warranty, the K2′s SPEAR operating system  optimizes flash endurance allowing Kaminario to offer a 7-year warranty. (SPEAR is Kaminario’s scale-out performance storage architecture operating system software.)

The original K2 was built with a focus on high availability and data protection and this new version of the K2 expands upon that.  Its self-healing features, automated data mirroring,  and hot swappable components have been configured to minimize performance impact upon recovery from a failure.  Space-efficient snapshots further assist in data recovery.

The new K2 has been optimized for mixed workloads and applications that can avail themselves of its consistently high read/write bandwidth and IOPS along with its low latency.  It also adds the iSCSI interface to the K2 platform, which previously only communicated through a Fibre Channel interface.

Kaminario says that the new K2 is fast to install and can be brought into a production environment in less than a day.  The system has been designed to support non-disruptive upgrades to minimize downtime and support costs.

What Kaminario is most proud of, though, is the product’s scalability. Although it has only qualified systems as large as 120TB, the company is confident that the system will successfully scale well beyond that level. The system has been designed to allow performance to linearly scale with all applications and, in particular, within virtual server and VDI environments.

Whereas HGST’s first-generation UltraStar SAS SSDs used SLC flash, the new SSDs are based on 25nm MLC flash but offer the same warranties as HGST’s prior generation.  Even so, performance for the new SSDs is significantly faster than that of their SLC-based predecessors, with no reduction in wear or lifetime specifications.

These SSDs are the first to support the 12Gb/s SAS interface, according to the company.  HGST’s 400 series of SSDs was limited to 550MB/s by the 6Gb/s SAS interface.  With the new interface, bandwidth and IOPS increase while latency decreases.  HGST has also emphasized consistent performance, which is a new focus for most SSD makers.  Performance is not only better than the prior generation but also has a narrower distribution, as is shown in this post’s graphic: a latency histogram.  The red columns are the new drive and the blue columns are HGST’s prior generation.

All models offer a 1,200 MB/s sequential read rate and 145K read IOPS.  Sequential writes are 750MB/s for the MH and 700MB/s for the MM and MR.  Write IOPS appear in the table.

The SSD Guy tips his hat to HGST for a job well done.

Objective Analysis has a report covering the enterprise SSD market.  The Enterprise SSD: Technologies and Markets can be purchased for immediate download from the Objective Analysis website.

Are these two announcements related?  Well, The SSD Guy thinks they are!

Higher-RPM HDDs help to accelerate disk accesses by a small percentage while a hybrid can boost speeds significantly.  According to Seagate, the new hybrids are: “Up to 5× faster than a traditional 5400-RPM HDD.”  A hybrid offers a bigger speed boost while reducing the power requirements, since faster spindle speeds eat up a lot of power.

Seagate introduced the new hybrid family in a multi-message press release:

• The new hybrids are endorsed by leading OEMs
• New third-generation 2.5″ hybrids are now available for mobile PCs
• Seagate is launching a 3.5″ Desktop hybrid

The new hybrids come in two diameters: 2.5″ for notebook PCs and 3.5″ for the desktop.  The 2.5″ drives are offered in two thicknesses – 7mm and 9.5mm.

The 7mm 2.5″ hybrid gets past a difficulty encountered by Seagate’s prior offerings.  The last generation of Seagate 2.5″ hybrids was only offered in a 9.5mm thickness which was too thick to find a home in today’s shrinking notebook PCs.  Intel’s Ultrabook specification calls for the entire PC to be only 13mm thick, which leaves a scant 3.5mm for the housing and keyboard on either side of a 9.5mm HDD or hybrid.  A 7mm version boosts that 3.5mm up to 6mm, which is a more manageable number.

The thicker 2.5″ hybrid ships in a 1 terabyte capacity while the thinner one is limited to 500GB.  A 2TB version of the 3.5″ drive has been announced.  All of these products have 8GB MLC NAND caches, twice the size of the 4GB SLC flash in the earlier Momentus XT.  The 2.5″ drives are based upon a 5,400RPM HDD while the 3.5″ product maintains the 7,200RPM speed of the older product.

Although this was not in the Seagate announcement, a Tech Report article reveals that the company has also upgraded its caching algorithms (which the company calls “Adaptive Memory”) to cache not only disk reads but to include writes as well, using a “Write Through” policy rather than the earlier “Write Around” version.  Cache writes are a thorny problem which require some caution to implement.  Since PC programs tend not to spawn many writes, Seagate was safe to put off write acceleration until the company was certain of its algorithms’ integrity.

I happen to be bullish on hybrids since they offer the high capacity and low prices of a standard HDD with speeds approaching those of an SSD.  Objective Analysis has published a report on this market: Are Hybrid Drives Finally Coming of Age? that can be purchased for immediate download from our website.

How sensitive are they?  Well, I have seen some overblown claims from SSD makers that shock will cause HDD head crashes.  I am not sure that I believe such claims, but I certainly do believe that an HDD’s actuator (the read/write head mechanism) can be shaken away from its track, causing a one-rotation delay before it can regain its position and access the desired sector.  After all, on a terabyte HDD the tracks are only about 100nm apart (that’s one ten-millionth of a meter)! It doesn’t take much vibration to move the head more than that much, so disk array designers pay a lot of attention to shock isolation to make sure that an actuator movement from one HDD is less likely to vibrate an adjacent HDD.  Even fans are a concern.

The graphic for this post is from a video that was made in Sun Microsystems’ Fishworks Lab way back in 2008.  It is only 2 minutes long and is well worth watching.  The engineer in the photo, Brendan Gregg, shouts his lungs out at a disk array, and the video shows a significant latency spike as a result.

Notice how loud the data center is to begin with, and you will wonder why vibration isn’t more of a problem even without the shouting.

In the data center this may not pose significant problems, especially if you don’t make a habit of shouting at the disk arrays, but in other environments, especially in industrial applications, this exercise makes it clear why many engineers choose to use an SSD rather than an HDD, even if they don’t need the the SSD’s speed.  I like to give the example of a jet fighter that shakes so much that the pilot may even worry about losing the fillings in his teeth.

(Thanks go to Tom Coughlin of Coughlin Associates for his help with the track widths, which he put as 300,000 tracks per inch.  Wow!)

The SSD Guy figured that Easter would be a good time to talk about these since everyone already has the Easter Bunny hopping through their minds!

It’s not hard to understand why the one-hop approach might be faster, but the two-hop approach offers pretty good performance for a very reasonable development cost, and that translates to a lower price.  While a one-hop controller requires a dedicated design effort, a two-hop PCIe SSD can be manufactured by almost anyone by teaming an off-the-shelf SSD controller with an off-the-shelf HBA or RAID controller, using firmware supplied by the manufacturers of the controllers.

Here is a list of the companies that I know of that are involved in either camp as of the writing of this post:

There will be more, and I have doubtlessly missed a few.  I will update this post as I am made aware of them.  I ask PCIe SSD makers not to be shy, but to contact me directly if I have missed them.

Objective Analysis covers the market for PCIe SSDs in its report The Enterprise SSD: Technologies and Markets, which can be purchased for immediate download from our website.

• The difference between the speed of system memory and that of a hard disk drive (HDD) is roughly 6 orders of magnitude, or 1 million times
• SSDs split the gap.  An SSD is about 1,000 times faster than an HDD, and is about 1,000 times slower than system memory.  Memory access times are measured in nanoseconds (ns), SSDs in microseconds (µs) and HDDs in milliseconds (ms)

The problem with understanding this (ns, µs, ms) is that to the most people these measurements are all so tiny that they appear to be about the same.

To put things in a more human perspective lets slow everything down by about 1 billion times:

• System memory access is now the same as a human heartbeat… Every heartbeat you get another piece of data
• An SSD access is about the time it takes you to walk a mile
• An HDD access is the amount of time that it would take you to bicycle from San Francisco to Miami

In these basic terms it is no wonder that everyone’s excited about SSDs.  Now you can walk to the neighborhood store for a carton of milk rather than bicycle across an entire continent.  Needless to say, the result is that you can get a lot more work done significantly faster.

It should then come as no surprise that I constantly repeat that SSDs should not be viewed as a replacement for HDDs, but as an element to be used sparingly in a system to accelerate performance.  SSDs will always be much faster and much more expensive than HDDs.  There will be a place for both technologies in computing systems for the foreseeable future.

The Halo storage software, which already boasts a rich feature set, has added a new API, a mobile access to performance information, and powerful analytics tools that track and report over 200 metrics in real time with unlimited scroll-back.

The Nimbus Gemini system has already been shipping for a couple of months and is finding acceptance in various markets: Analytics, Virtualization, Cloud Computing, and Scientific.  The largest Gemini 2U system, which Nimbus claims has the greatest raw capacity per rack unit, weighs in at 48TB raw, 336TB compressed, and can handle up to 1 million virtual machines.

The SSD Guy was impressed by the graphic for this post, which compares the space requirements for a 5 million IOPS Gemini against those of a competing production system.  (Click on the picture for a larger rendition.)  The chart may look just like the one that every single solid state storage vendor uses, the one that compares their SSD-based system against a competing HDD-based system, but in this example Nimbus is comparing its own system against a competitor’s flash-based system!

Nimbus claims to be the first storage system, either SSD-based or HDD-based, to use the new 16Gb/s Fibre Channel interface.  (They tell me that all of the 16Gb FC interfaces shipped to date are on the server side rather than on the storage side.)  The company adopts fast interfaces early to assure that the internal bandwidth of their flash is not throttled by insufficient network bandwidth.  Nimbus doesn’t want the system bottleneck to simply move from storage to the network.

It’s a pretty impressive announcement from an impressive company.  I’ll be interested to see what Nimbus achieves as the company progresses.