June, a busy month for Data Direct Networks.

more

Last month Data Direct Networks (DDN) made two major announcements, one on June 22nd and the second on June 23rd. The strategy of two major announcements on consecutive days is a bit questionable, not to mention the joint BlueArc/DDN announcement also on the 23rd. I think DDN did themselves a great disservice by making these two announcements on consecutive days. But no one asked for my opinion.

However before talking about the announcements how about some interesting factoids DDN offered at the pre-announcement briefings.

• Over 300% CAGR over the past five years.
• Seven straight profitable years with current annual revenues exceeding $130M.
• Over 90PB Shipped in 2008.
• They are installed on eight of the world’s top ten high performance computing systems.
• Claim over 450 post production and broadcast customers.
• Their customer Microsoft, Xbox Live Community supports 10,000,000 online users and Shutterfly manages 3Trillion individual photographs with 15PBs of DDN storage.

Ok, back to the announcements. The first was on 6/22 and established their presence in the Cloud with their WOS (Web Object Server). I covered this announcement in an earlier post. The second announcement on 6/23 presented their next generation platform architecture called the Storage Fusion Architecture (SFA) with the incumbent architecture the S2A is still available on the 9900 and 9700 product lines. The SFA announcement did not appear to enjoy as much fanfare so for those who missed it let me share some detail.

The SFA is a highly dense, high performance scale-up storage architecture. It targets the evolving storage requirements for large scale, multi-thread simultaneous file access that is characteristic of today’s WEB 2.0 and Internet workloads. These workloads are unpredictable, with queries of variable complexity, have access patterns that are increasingly random, have a high data access requirement, which means a high IOPS, and with a need to deliver large as well as small files, high bandwidth is a requisite.

The active /active architecture can sustain 300k of random disk IOPS and can achieve a burst (cache) performance number of 1M IOPS. Bandwidth is quoted at 10GB/s, read and write. Each system supports a maximum of 1200 drives in two standard racks and assuming 2TB drives, that is a system capacity of 2.4PB or 146.88GB/sq ft, the highest density configuration for online storage in the market. Even when using 1TB drives DDN beats out most other online mass storage options with their unique, dense packaging.

Connectivity options are either 16x8Gb/s Fibre Channel ports or 8x40Gb/s Infiniband ports. Internally, a 480Gb/s SAS data network has been implemented to supply sufficient headroom to meet all the array housekeeping requirements without impacting the performance of an application.

Unique to SFA is its support SAS and SATA hard drives (2TB now supported, SFA only) and SSD all within a single system. Again SSD is supported on SFA only. Obviously the intent is to provide the physical storage tiering hierarchy but what I find surprising is the lack of any integrated tools to enable users to employ tiering; a bit like a cell phone without the SIM card. There are many storage companies who have the physical storage tiers available in their solutions but very few who have a meaningful automation of the feature. Simple to use, meaningful automation is what delivers the benefit of tiering to the user. This shortcoming is probably the biggest weakness I perceive in this otherwise impressive announcement.

Serverless computing; unique approach to server consolidation.

more

First question, what is serverless computing?

Simply, the co-hosting of processing, storage and I/O onto a single device. The end user benefit being the elimination of dedicated application servers by co-opting excess compute power that reside in many storage servers to host compute-intensive workloads.

The company that has taken the lead with this common sense idea is start-up Pivot 3. It is not a concept that is universally applicable as pointed out by Lee Caswell, Founder and CMO “This technology is ideal for environments with large amounts of storage and relatively few applications, such as surveillance, medical imaging, video streaming and data de-duplication," Pivot 3 has focused on exploiting this unique server virtualization approach to the video surveillance market.

Interesting to note that the video storage surveillance market has been estimated by Frost and Sullivan to grow to $3B in 2013 and by their estimates savings as a result of serverless computing will be in the order of 7.3%.

So how does Pivot3 technology make this idea work? While tradition approaches to SAN hardware is based on proprietary hardware Pivot3 software enables commodity x86 storage servers and server virtualization software to concurrently run its storage virtualization software across multiple appliances with one additional server application per appliance. As with conventional server virtualization, server hardware is eliminated but unlike conventional virtualization, storage appliances are limited to host only one additional application with each application have access to the entire clustered storage array's bandwidth and capacity.

This unique server consolidation approach reduces not only hardware but power consumption, heat generation and rack space.

Being some one who is interested in green IT and admittedly it does not have universal application but this is one of the most pragmatic options I have yet uncovered and one that should get the attention of IT professionals.

Less is always best.

With the Mercury™ Series, BlueArc moves out of their comfort zone and targets the midrange commercial market.

more

BlueArc announced today the availability of their next generation of Network Storage Solutions, the BlueArc Mercury™ Server.

Targeted at the mid-range storage market this downward extension of the Titan portfolio gives BlueArc an entry into what their VP of Marketing, Bridget Warwick, calls the enterprise mid-range storage market.

Storage consolidation is the lead message that underpins this announcement with their go-to-market strategy promoting the elimination of storage silos, reducing infrastructure and operating costs, all with the end goal of reducing overall storage TCO. Supporting this strategy is Intelligent Tiered Storage a feature that BlueArc looks at as one of their major competitive advantages. Storage tiering is one of today’s more topical subjects and if the BlueArc technology lives up to its promise of effectively automating storage tiering then this feature will most certainly appeal to the thoughtful storage administrator as they struggle to improve their overall storage efficiency and operational costs. Many vendors are talking about tiering but few (any?) have effectively automated the process.

BlueArc has positioned Mercury as a cost effective consolidation solution that addresses the data island sprawl that commonly permeates growing enterprises. However, I am not sufficiently comfortable in my knowledge of their pricing strategy to comment on the pragmatism of this approach.

However, BluArc has missed a great opportunity to promote what I perceive to be a powerful unified storage message. BlueArc is data type agnostic supporting simultaneous access to file data and block I/O through native support of NFS, CIFS and iSCSI. They do talk about supporting a wide range of third party hardware platforms with particularly tight relationships with HDS and Data Direct Network. Being data type agnostic, having the ability to aggregate disparate heterogeneous storage silos and present a single consolidated management view is a powerful consolidation message not exploited.

The heart of BlueArc hardware platforms is the hardware accelerated file system, SiliconFS™. Accelerated data operations is the benefit delivered by in effect turbo charging the file system through a combination of field programmable gate array’s (FPGA) and multi-core processors. The BlueArc architects select which functionality will deliver the most benefit from hardware acceleration and have the flexibility to adjust this selection on the fly. Worth noting that the use of FPGA tends to avoid the design lag normally associated with proprietary silicon and should give BlueArc a design nimbleness that should translate to a competitive advantage.

This object based file system and the hybrid hardware architecture delivers what BlueArc consider to be their “unfair advantage”, rapid development response time, transparent data mobility, advanced data protection and a rich storage virtualization engine.

The technology that is Mercury™ is the same as that used in Titan simply downsized to fit a mid-range price point. Mercury™ will be initially introduced as the model 50 with the model 100 to follow. I detect a potential position problem with the Mercury 100 sitting on top of the Titan 3100; relative price points I suspect will be an interesting discussion. Observing how BlueArc manages this positioning challenge will be interesting.

The Mercury Series gives BlueArc an opportunity to expand their indirect channels. They expect to develop their reseller channel and to increase the business opportunity with their one OEM, Hitachi Data Systems. Mercury will also be sold through the direct BlueArc sales team. Caution, be wary of developing channel conflict.

BlueArc are moving into unchartered waters with this push into the commercial and highly competitive mid-range space and look at this move to quadruple the market available to them. However, if my assumption is correct and they will price at the upper end of the mid-range price point it remains to be seen how much of the market will be willing to pay the price premium BlueArc represents. This challenge speaks to the somewhat daunting task of changing the perception the market has of BlueArc with messaging, awareness and conversion proof points that resonate with this new prospect pool.

There is no question that BlueArc has outstanding technology but the mid-range commercial space takes them out of their HPC comfort zone where price tends to be less influential than performance and where the buying motion is less mercenary.

Buyer beware when vendors talk Storage Utilization.

more

Words games are not the sole prerogative of the politician. Our marketing colleagues have a tendency to push the envelope with the latest example being some folk’s interpretation of the term Storage Utilization.

When I think of storage utilization I tend to think of the capacity available to the user to perform useful work after all system overhead has been considered. I do not consider the capacity required to protect or manage stored data as user available storage. So when data is being protected with mirroring for example, storage utilization is significantly less than when RAID 5 or RAID 6 is implemented.

Date Protection Scheme...................................Storage Penalty Mirrored...................................................................50%
RAID 5......................................................................17% (5+1 Raid Group)
RAID 6......................................................................33% (4+2 Raid Group)

Except in the case of mirroring the bigger the RAID group the smaller the penalty which is one of the configuration compromises, now add the system OS etc, you get the point.

So when you see eye catching guarantees of Storage Utilization make sure you understand how this utilization is being defined. You may be surprised.

Server “Wake on LAN” technology enables Xiotech’s PowerNap energy efficiency feature.

more

With their recent announcement of PowerNAP™, Xiotech joined the ranks of storage vendors offering energy efficient storage. Although currently restricted to the Emprise 5000 I suspect that it will permeate upwards in the not to distant future.PowerNAP™ is an uncomplicated approach to energy efficiency. Storage arrays are powered down based on a predetermined schedule. Think of storage in a remote office with well defined hours of business, a back-up or DR target or an inactive archive data repository. These use cases all have predictable time when the hosted data is unlikely to be requested and instead of simply allowing the inactive disks to continuously and unproductively spin, the subsystem is powered down.Wake on Lan originated in the server world; simply put it activates sleeping devices by sending out what is called a Magic Packet, a broadcast frame that contains a 6 bytes of 1’s followed by 16 repetitions of the target devices MAC address. As some organizations may not permit broadcasting or unicasting packets across a LAN it may take some “buddying-up” by the storage folks to their networking compadries to get the required support.This approach is not as dynamic as some other power management features where it is the activity profile of the data that determines the power state of the drives. While WOL can respond to unpredictable circumstances and wake the drives up, out of sequence power down required a manual intervention.How much of a negative is this approach. I guess it depends. If workloads are highly predictable, as per the examples above then this approach should work fine. However this is an all or nothing approach with the array powered up or powered down. It lacks the granularity or flexibility to optimize energy efficiency in systems that host a broad spectrum of active and inactive data. Automating the process is via scripts and while not particularly onerous its ease of use falls short of competitive GUI driven management. Admittedly my impressions are based on documentation and not hands experience but not automating data and device integrity checking is questionable at best.On the plus side when the system is powered down only consumes 1W, impressive! On power down all I/O’s in progress are completed and the cache is flushed prior to shut down. On receiving a WOL request the array is up and ready to service data after an acceptable 60 second delay.Xiotech are not alone in using the WOL approach for storage solutions. At SNW I looked at the NAS boxes from Netgear. These compact 1.5TB to 6TB storage units targeted at SMB, professional office or even the power geek’s home installation.

Innovative start-up makes clustered storage affordable for SMB – gets funding from the State of Indiana.

more

Scale Computing, a San Carlos California start-up and soon to be headquartered in Indianapolis, Indiana recently received a public/private, $5M funding injection, with a $2M grant coming from Indiana’s 21 Century Research and Technology Fund and the balance from private investors led by the Blue Chip Venture Group.

Scale Computing is a small aggressive organization that is still working out of the proverbial “garage” in San Carlos, but that is all about to change. Their humble surroundings however have not prevented them from developing an innovative clustered storage solution which has all the attributes that resonate with SMB’s looking for performance but affordable storage solutions and those challenged by dedicated data intensive application. Not to mention creating market indigestion for the more established players.

The Scale Computing clustered architecture is built on commodity components bringing a cost and flexibility advantage not enjoyed by more tradition designs. Each 1U node is based on a Supermicro server and comes in 1TB (SN1000) and 2TB (SN 2000) configurations. TrueCluster™ is the storage cluster management system and is a tight integration of licensed intellectual property, open source software and Scale Computing’s proprietary technology. The TrueCluster™ architecture automates storage services that reduces management complexity, supports what is called predictive hardware failure technology and enables a powerful clustered file system. All nodes are managed through a UI which having recently watched a live demonstration, justifies my describing it as intuitively simple.

The foundation of their clustered file system is IBM’s GPFS which has already earned considerable credentials in the HPC world but now has been tamed by Scales innovative engineering team to deliver not just its powerful functionality but with the ease of use that enables it as a practical option for the commercial world, particularly the SMB and non-sophisticated storage users.

Some highlights of the Scale solution are:

• “Pay as you grow” Storage. Addresses the questionable practice of paying for storage you do not yet need, a waste of CapEx and OpEx dollars. A basic attribute of clustered architectures is scalability which Scale Computing has enhanced with a highly granular approach that starts with a minimum configuration of 3TB and grows in increments of 1TB of usable capacity. These growth chunks realistically match the growth need of the smaller user. Clusters can be expanded to a single file system capacity of 2.2PB.
• TrueCluster™ is a symmetrical clustered solution. This means there is a peer relationship between all nodes and with each being equal each has the capability to change and control the entire cluster. This elimination of any “master” node eliminates a single point of failure and delivers a robust, highly available solution normally associated with enterprise data centers.
• Unified storage: This means a storage system that can be used with file systems and storing file data; or can use an application that does block-based I/O (database/email) right to that storage system. TruCluster can serve block and file based data from the same cluster via iSCSI, NFS or CIFS. An interesting consolidation option which previously was only available on much more expensive solutions.
• Commodity based hardware. Ride the declining commodity curve and the dynamic innovation curve to drive competitive pricing and rapid integration of new technology driven options.
• Disruptive pricing: A clustered storage solution that has an entry price point under $13k (for minimum configuration of 3xSN1000) and a per “usable TB” cost that drops to $2500 (when purchasing the SN2000) will certainly create indigestion for the current vendor establishment.
• Ease of Implementation, Management and Use: Innovative use of artificial intelligence technology to eliminate complexity and simplify the customer experience, including predictive self healing attributes, are considered by Scale to be a significant competitive advantage. These plus the scalability and flexibility inherent is a clustered architecture are attributes that strongly resonate with not just the SMB user but all storage practitioners.

As noted Scale Computing is a start up organization that has all the enthusiasm, expertise, energy and “moxy” you would expect from a team of highly motivated and bright individuals. The question is their ability to execute in the face of commercial reality and to navigate the hurdles they will undoubtedly encounter as their customer base grows, not to mention the increased competition that will develop as they start to appear on the radar screens of the established players in the storage market.

They are not without success with a strong reseller ecosystem developing with such resellers as Ryjac Solutions and MicroAge and customers in healthcare, education, legal and the construction verticals and with a win at Motorola they have now registered their first fortune listed customer. The first beta unit was shipped last November with GA units being shipped in February of 2009. Approximate unit shipments to date, is close to 150.

So far, so good! However, they are moving into a new phase of their corporate and market development but the strength I perceive in their core value proposition and their past entrepreneurial success, would suggest that they have a better than average chance of success.

Is NetApp missing a golden green opportunity?

more

NetApp’s customer facing green strategy or lack thereof was one of my few disappointments at the recent NetApp analyst day. At the end of the sessions I was left with the sense that this was not on their radar and that I have some difficulty in accepting.

However, Green IT was not totally ignored. Dave Robbins, CTO, IT Infrastructure, gave an excellent overview of NetApp Green IT activities from an IT practitioner’s perspective and his efforts and those of his team are to be applauded.

But what about NetApp’s customer facing green strategy? If there is one it is not obvious. In the product presentations there was no mention of how NetApp is driving environmental and energy efficiency even when discussing data reduction technologies. Although I am a green advocate I am also the first to admit that green per se does not necessarily win business but energy costs, space efficiencies, ewaste management etc are increasing becoming relevant even for the pragmatist whose sole worry is operational costs. Even without the pending legislation Washington has in store for us a strong, well articulated green strategy is a valuable sales asset and its importance will continue to increase with the proliferation of worldwide environmental legislation. EMC, Dell, HP and IBM have all received and understand this message.

Let’s look at what Dave’s team has accomplished.

By implementing VMware they reduced from 4600, x86 based clients to 230 saving $1.3M in power and cooling and reduced the number of racks needed by 182. Admittedly not driven by NetApp technology, but a heck of a story.

How about their Oracle environment? They consolidated midrange servers to logically partitioned enterprise class servers which added 300% capacity to production servers and increased the development landscape by 143%. Replication time was also shortened by 2400% (from 1-2 days to 1hr) and the need to for a 74% increase in floor space was avoided as was the need for 81% additional power which saved an additional draw on the power grid of 236kW. Again not a directly related to NetApp technology but another great story and a great illustration of what a bit of planning can do.

So what was achieved as a direct result of NetApp technology? Primarily a consolidation/virtualization strategy that reduced storage systems from 50 to 10. This freed up 19.5 racks in the data center and reduced power consumption by 41,184KWh/mth which reduced energy consumption by 32%. Air conditioning was reduced by 94 tons which translates to eliminating approximately 1M lbs of CO2 annually and as a bonus they also increased storage utilization by 60%.

For the life of me I do not understand why NetApp are not exploiting the great work being done by Robbins and crew particularly as they are showcasing NetApp technology. In today’s popular parlance they are eating there own dog food. The NetApp folks have established solid green credentials in the industry, they were involved with the Data Center Energy Forecast Report sponsored by Accenture, Silicon Valley Leadership Group, DOE and many others, in short they have a thought leadership position in Green IT. It would be a shame to let that golden opportunity go to waste.