Storage tiering, where data is automatically placed and migrated between different storage media, improves the performance of a system by exploiting the access characteristics of data. However, the net effect of tiering sometimes gets overlooked in discussions about maximizing storage system performance.
• Increased utilization by reducing the number of storage systems with reserved or unused capacity.
• Reduced requirements for management and administration of storage systems. Along with bringing more advanced management software with newer storage systems, reducing the number of storage systems reduces the amount of time required for administration.
• Reduction in power, cooling, and physical space is a common result of implementing new technology. Consolidating systems where a new storage system can support larger workloads typically has a greater impact on the environmental reductions.
• Reduced maintenance costs/support contracts from fewer storage systems.
What role does storage tiering play in consolidation? Tiering can maximize the performance of a storage system and may be the most important enabler for consolidation from an economic standpoint.
Implementing storage tiering on solid state drives (SSDs) and high capacity disks requires capabilities built into the embedded storage system software to intelligently and automatically move data for optimal performance.
Results from deployments in customer environments verify the effectiveness of storage tiering, even when SSDs make up only four percent or less of the total capacity. This brings a new economic calculation to bear for storage tiering and the return on investment for a storage consolidation project.
Vendors are increasingly focusing on the automation and effectiveness of the tiering implementation, especially with the emergence of SSD as a storage tier. These are not esoteric elements in a storage system but critical, high value functions with the potential for storage efficiency improvements and relatively quick economic payback. This means the understanding of how the tiering works, how effective it is, and the differences in product costs with the amount of SSDs required for optimization requires evaluation and independent information for making a decision. Tiering has huge payback and needs to be included in the strategy for IT operations.
(Randy Kerns is Senior Strategist at Evaluator Group, an IT analyst firm).
Object storage has been hailed as a better way to store files than traditional NAS, and perhaps even a long-term replacement for file storage. Now open-source software vendor Gluster has integrated object and NAS capabilities in the same file system.
The GlusterFS 3.3 beta that became available this week lets users store and access the same data as an object and a file. They can store objects and then access those objects as files, or view and access files as objects. The idea is to make it easier to migrate file-based applications to object storage so they can be used in the cloud.
GlusterFS is an open source file system that scales to petabytes with global namespace. Version 3.3 has an object interface integrated into that file system.
“Probably 95 percent of enterprise applications haven’t been able to leverage object storage and move to the cloud,” Gluster director of product marketing Tom Trainer said. “Integrating it in one system will accelerate the integration to object storage.
Gluster customers can access data as objects from any Amazon S3 compatible interface and access files from its NFS and CIFS interfaces. Trainer said service providers can use GlusterFS to build Amazon-like storage for customers. It can also be used to migrate legacy applications to the cloud and scale NAS across the Internet to a public cloud.
There have been other approaches to integrating object and file storage, although Trainer maintains that GlusterFS has the deepest integration of object and file interfaces.
Most object storage products such as EMC Atmos, Scality, OpenStack and Amazon S3 don’t have file systems. Caringo, which began as object storage, added an NFS and CIFS interface for its object store and Nirvanix’s CloudNAS makes its object storage look like NAS to an application.
EMC’s earnings today revealed few surprises with revenue of $4.85 billion and income of $793 million at or slightly above expectations, but the results include a few interesting product trends:
• Isilon scale-out NAS is EMC’s fastest growing software platform, doubling in revenue over Isilon’s performance as a standalone company last year. EMC said its other Big Data products — Atmos and Greenplum — also doubled revenue year-over-year, but they were from smaller bases. EMC acquired Isilon for $2.25 billion last November. Isilon’s product revenue for the second quarter of 2010 was $64.9 million.
• More flash solid-state capacity shipped with VMAX and unified storage systems in the first half of this year than in all 2010.
• FAST VP automated tiering is EMC’s fastest selling storage software product, and more than 90% of systems with FAST are shipped with flash and SATA drives.
• Vblock integrated stacks sold by the VCE partnership between EMC, Cisco and VMware had more revenue for the first half of 2011 than in all of 2010.
• EMC took a charge worth $66 million – lowering earnings per share by 2.5 cents – for “remediation” to retain customer loyalty related to a breach of RSA’s security tokens at defense contractor Lockheed Martin.
• The Information Intelligence Group, which includes Documentum and EMC’s compliance products and services, had a 5.1% drop in revenue from last year. EMC CEO Joe Tucci said the vendor remains behind its IIG products and is determined to get results back on track. “We’re going to stick with it,” he said. “We’re convinced that by the end of this year or early next year, we’ll return this business to growth.”
Dell’s acquisition of Ethernet switch maker Force10 today should end the expectations that Dell will buy Brocade, which sells Fibre Channel and Ethernet gear.
Financial analysts have speculated and even prodded Dell to acquire Brocade as it tries to move up into an enterprise class vendor instead of a PC and server specialist. There have been whispers for at least a year that Dell was considering Brocade, but it instead followed Hewlett-Packard’s lead in going just for Ethernet switching. HP also looked at Brocade before buying 3Comm in 2009.
Apparently, system/storage vendors prefer to own their Ethernet technology while getting FC connectivity from Brocade and Cisco. The move to Ethernet for Dell and HP is motivated at least in part by Ethernet switch market leader Cisco’s getting into the server business with its Unified Computing System (UCS).
Price may also have been a factor for Dell. It did not disclose how much it will pay for Force10, but it is believed to be around $600 million to $800 million while Brocade would command billions of dollars.
Passing on Brocade doesn’t mean Dell won’t buy another storage company, though. It has picked up EqualLogic, Exanet, Ocarina and Compellent since 2008, and Dell executives including Michael Dell see storage as a major part of the company’s future.
All new storage technology doesn’t come from startups, although you might get that impression by reading about industry acquisitions.
The reasons most often listed for acquisition of a start-up company are:
• Technical infusion (technology acquisition)
• Expansion into a new business area (new technology and staff)
• Complementary solutions (filling in a product line hole).
These reasons would lead to the conclusion that startups bring new technology to customers more effectively than large established companies. Considering that popular technologies such as data deduplication, thin provisioning and iSCSI storage were originally brought to market by startups, there is merit to this line of thinking. But it is not such a simple issue. Large vendors do have brilliant and dedicated people, but developing and bringing a product to market in these companies can be a complicated process. That’s because they create a corporate structure that often makes it difficult to take a new idea or approach and bring it to reality.
Large companies have processes that their people are required to follow, making it difficult to innovate. Any initiative or idea must conform to their interpretation of the company process, and there are organizations and people inside each company that can create enough resistance to hinder realization of the new ideas. I call these people and processes the Department of Revenue Prevention.
If a large company has an entrenched Department of Revenue Prevention, it is easier for people with ideas to take them through the startup route. That route has less resistance, and innovators’ time and efforts are not spent battling the department but actually moving the innovation to market. Unfortunately, the rewards may be limited based on what must be given up to get the funding necessary to take the technology innovation to a product stage. Ultimately a startup may not be successful for a variety of reasons, including:
• A bad board assigned by investors that do not understand the market, the technology, or what is required to bring the technology to fruition
• Missing or subpar key people in areas such as strategy, marketing, and sales.
• Technology that may not meet customer needs at the right time -– either being too early or too late.
Large companies that understand how to nurture and develop the ideas of their talented people will be more successful than those that succumb to the bureaucratic sprawl and paralyzing Department of Revenue Prevention structure. Even inside these great companies, things change over time and bureaucracy spreads. To re-invigorate a company requires periodic review and change to enable innovation. It’s either that, or continue to acquire other people’s ideas.
(Randy Kerns is Senior Strategist at Evaluator Group, an IT analyst firm).
While EMC formally launched its VMAXe enterprise storage system to compete with IBM’s XIV (as well as Hewlett-Packard’s 3PAR) this week, IBM was giving XIV an overhaul.
IBM launched what it calls XIV Gen 3 with InfiniBand connectivity between modules, 2.4 Ghz quad core Nehalem CPUs, 2 TB native SAS disk, and 8 Gbps Fibre Channel support. By next year, IBM also expects to offer up to 500 GB of solid-state drive (SSD) capacity per module for a total of 7.5 TB in a fully configured 15-module configuration. According to senior management consultant for IBM system storage Tony Pearson’s Inside System Storage blog, XIV will use SSDs as DRAM cache similarly to NetApp’s Performance Accelerator Modules (PAM) – a product IBM resells as its N Series.
None of the XIV enhancements are ground-breaking, but IBM claims to get a two- to four-times boost over Gen 2 for workloads such as transaction processing, sequential reads and writes, and file and print services, and applications such as Microsoft Exchange and Hyper-V, Oracle Data Warehouse, and SAS Analytics Reports.
IBM will keep XIV Gen 2 around for at least a year for customers who don’t need the new system’s performance or capacity (Gen 2 uses 1 TB drives).
In case you’re wondering, Gen 2 was the first version of the product that IBM launched in Sept. 2008 after acquiring XIV the previous January. Gen 2 had different disks, controllers, interconnects and software enhancements over the Gen 1 product that it bought from XIV.
While IBM characterized XIV as a Web 2.0 system when if first purchased it -– the same label EMC used to describe it during the VMAXe launch –- Pearson wrote that XIV is a full-blown enterprise system that competes with EMC’s high-end VMAX. “As if I haven’t said this enough times already, the IBM XIV is a Tier-1, high-end, enterprise-class disk storage system, optimized for use with mission critical workloads on Linux, UNIX and Windows operating systems, and is the ideal cost-effective replacement for EMC Symmetrix VMAX, HDS USP-V and VSP, and HP P9000 series disk systems,” Pearson wrote.
He did point out, though, that the DS8000 remains IBM’s platform for mainframe connectivity.
The XIV launch was low-key and took second fiddle to Big Blue’s IBM zEnterprise 114 Mainframe Server rollout this week, as Enterprise Strategy Group analyst Mark Peters pointed out on his The Business of Storage blog. Peters was generally impressed by the new XIV, though.
“The third generation of XIV is all about adding performance– and plenty of it,” Peters wrote. “Besides more cache, more/faster ports, and a change to SAS drives, there’s also Infiniband connectivity within the XIV (helping, surprise surprise, with performance) and ‘spare’ CPU and DRAM slots for ‘future software enhancements’ … IBM is keen to point out that the SSD is transparent caching, with no tiers per se to manage. Of course, it would be, since XIV has always proclaimed there’s no need to tier. But, pragmatically, as a user I’d only worry if it economically makes the system better and still does it without me needing to manage things. Assuming so, then I’ll give it a thumbs up and leave the semantic debate to others.”
Nimble Storage today added a smaller model of its combination primary storage/backup platform and $25 million in fresh funding.
Nimble launched the CS210, a year after it came out of stealth with CS220 and CS240 systems that combine iSCSI, integrated inline compression and replication to optimize and protect data, and flash to accelerate performance. The startup also said Artis Capital Management has led its fourth funding round, bringing its total funding to $58 million.
The CS210 is an entry level version of the Nimble platform. The CS210 has 8 TB of usable capacity, and costs $38,000. Comparatively, the CS220 has 16 usable TB for about $58,000 and the CS240 holds 32 TB for about $88,000.
Besides capacity, the difference with the CS210 is that it doesn’t support 10-Gigabit Ethernet (GbE) out of the gate. The CS210 comes with four GbE ports, while the CS220 and CS240 have either six GbE or two 10 GbE ports. All the systems use 1 TB or 2 TB 7200 RPM SATA drives with up to 1 TB of multi-level cell (MLC) flash in 100 GB drives. The systems cache copies of hot data blocks in flash and send it off to the SATA disk.
Nimble marketing VP Dan Leary said the vendor has more than 100 customers in three full quarters of shipping systems. He said most customers use it for primary applications such as VMware and Microsoft Exchange and SQL while larger companies use it for test/dev and other specific applications.
Leary said he expects the CS210 to appeal to organizations “a little bigger than the classic SMB. We see it fitting at the low end of the mid-market for companies that want a primary system or a remote/branch office of a company that might have a CS220 or CS240 already. It’s also a good fit for customers replicating to a DR facility who have 90 days of snapshots at their main site but can get away with only 30 days in the DR site.”
Nimble’s latest funding closely follows the $16 million round it announced last December. Leary said the startup hasn’t burned through that previous funding haul, but benefitted from an attractive valuation from new investor Artis Capital Management. Artis was a major shareholder in Data Domain before EMC acquired the deduplication backup specialist.
Leary said Nimble will use the funding to expand sales to Europe and Asia. The company has 80 employees today and no sales team outside of North America.
Nimble switched CEOs in March, hiring former NetApp executive and Omneon CEO Suresh Vasudevan to replace founder Varun Mehta, who remains with Nimble as VP of Engineering and sits on the board.
Storage played a big part in VMware’s vSphere 5 launch Tuesday, as the vendor introduced a new software product called vSphere Storage Appliance and made enhancements in the areas of storage management and provisioning, replication and disaster recovery in virtual environments.
“Storage plays a central part in what we’re doing [with vSphere 5],” VMware senior product marketing manager Mike Adams said. “A lot of it has to do with advancing the cloud, but we’re also trying to help people become more efficient with storage.”
vSphere Storage Appliance is for SMBs, and lets them turn internal disks into shared storage that is required to reap the benefits of vSphere. Customers load the software onto a server and can point it at one or two additional ESXi targets to create a storage pool – similar to products such as Hewlett-Packard’s Virtual Storage Appliance (VSA). The appliance will cost $5,995 as a standalone product and $7,995 as part of a bundle with vSphere 5.
The first version is limited to three servers. “This is for SMB customers who can’t afford or don’t have the know-how to set up a SAN,” Adams said. “They can use vMotion for live migration and VMware HA for failover of virtual machines. They both require shared storage.”
Other new storage features from the vSphere rollout included:
Storage Distributed Resource Scheduler (DRS). This extends the DRS feature from the compute side to storage, helping customers quickly provision virtual machines to storage pools. DRS takes advantage of new vStorage APIs for Storage Awareness (VASA) to place data and load balance based on I/O and available capacity.
Profile-Driven Storage. This lets users map VMs to storage levels based on service level, availability or performance needs. VMs running applications that need the highest performance can be mapped to tier one storage, with less critical apps mapped to lower tiers. Customers associate tiers with service levels for performance and available capacity.
vSphere Replication. Building replication into Site Recover Manager (SRM) 5 removes the need for array-based replication, allowing customers to replicate data between different types of storage systems. It also adds automated failback and planned migration between data center capabilities. And while VMware presenters didn’t talk much about it during the public launch, the vendor also rewrote the code for its VMware HA DR product.
VMware vStorage APIs for Array Integration (VAAI) support for thin provisioning, NAS hardware acceleration. vSphere will inform arrays when files are deleted or moved by Storage vMotion so the space can be reclaimed. It also monitors capacity on thin provisioned LUNS and warns users when they are running out of physical space to avoid oversubscription with thin provisioning. The new hardware acceleration for NAS includes a full file clone that enables the NAS device to clone virtual disks, similar to the VMware’s full copy feature for block arrays. It also has a thick virtual disk feature that lets administrators reserve space for an entire virtual disk. Previous versions of vSphere always created a virtual disk as a thin provisioned disk.
vSphere 5 also adds NFS support for its Storage I/O Control feature that prioritizes I/O of virtual machines in shared storage to reduce latency.
When talking to CIOs, IT directors and managers, I’m sometimes surprised by what they know – and don’t know – about industry developments. During an education session I held recently, the IT people told me they had not heard of VAAI, the VMware vStorage APIs for Array Integration. This surprised me, given the performance gains yielded by VAAI with storage systems that supported them.
I explained VAAI and the improvements from using vSphere 4.1 and storage systems with VAAI, and then inquired about why they had not heard about it.
Most said that they did not have time to research information themselves and looked to “trusted advisors” for that type of information. The trusted advisors could be a small set of salesmen or sales engineers or the people from well-known independent firms. If the information hadn’t been pushed to them in sessions such as the one I was conducting, they not might not hear about it.
Digging further into their reluctance to research information, there was a general feeling that much of the written information they received had so much hyperbole (note: their actual word was “BS”) that the facts and useful information were obscured.
This means that much of the vendor marketing “amplification” was actually a detriment instead of an effective method of relating the virtues of a product or company. The way IT directors and managers look at this information has big implications in both marketing and in the delivery of storage technology education.
Obviously this is just a personal observation and not a scientific study (as opposed to a marketing study that is intended to obtain the desired results). But it does indicate that any company that invests in delivering information on storage products and technologies should evaluate the effectiveness of its messaging. It also shows the trusted advisor remains the best means of communication.
(Randy Kerns is Senior Strategist at Evaluator Group, an IT analyst firm).
Scale-out NAS is becoming popular, with most major vendors offering these types of products. As a reminder, scale-out NAS systems will increase performance and capacity at the same time – although you don’t have to scale the systems in the same ratio. You can add controllers for performance, storage for capacity, or both.
There are several things to look at when considering a scale-out NAS system.
1. Is a single namespace provided across all the nodes (also called controllers or heads) so that a file system can be spread across the nodes but the user does not need to take any special action for accessing a file? There are different ways that a single namespace can be implemented, and some may be better than others. Mounting or sharing a file system on a scale-out NAS system should require no more effort than if it was on a single-node system.
2. Does the management software manage across all nodes as an aggregate but still allow individual node communication to detect problems in the system?
3. Is there load balancing across nodes? The load balancing can be automatic when files are stored to distribute data across the different nodes. Will data automatically be redistributed across nodes (in the background) for capacity or load balancing?
4. Can it scale independently? In other words, can you scale nodes for more performance and the underlying storage for more capacity? This provides the greatest flexibility in usage. If the answer is yes, then how many nodes can the system scale to include? And how much capacity (including storage controllers) can it scale to?
5. Is there a back channel for communication between nodes? This requires another communication path between nodes rather than using the same path clients may be using to access data. Examples of this may be an InfiniBand connection between nodes or a 10-Gigabit Ethernet connection. Usually there would be a pair of back channels for availability.
6. Are there any features that are not included that would normally be part of standard NAS systems? A few to consider are: snapshots, remote replication, NDMP support, NFS and native CIFS support, security controls such as Active Directory, LDAP and file locking for shared access between CIFS and NFS, anti-virus software support and quotas.
7. Does the scale-out NAS support both small and large files? Some of the distributed file systems used for scale-out NAS come from the high-performance computing area where the optimization was around large files. It is important to understand whether the system supports small files and large files equally.
This list is a first level look with more detailed differences to be explained in an upcoming Evaluator Group article.
(Randy Kerns is Senior Strategist at Evaluator Group, an IT analyst firm).