Friday, January 29, 2010

Unified fabrics: 10GbE iSCSI vs. FCoE

January 29, 2010 -- The consensus is that 10 Gigabit Ethernet (GbE) will be the basis for unified fabrics/converged networks, but which storage protocol will become the de facto standard for SAN/LAN convergence?

In a recent study "Benefits Of SAN/LAN Convergence," conducted by Forrester Research and commissioned by Dell, 10GbE iSCSI ranks highest among users as the protocol of choice for unified fabrics.

Forrester conducted an online survey with 213 storage professionals in the US, UK, China, and the Netherlands, and 10 in-depth interviews of the same audience. The results show that more than double the respondents selected 10GbE iSCSI (56%) versus the next closest choice, Fibre Channel over Ethernet (FCoE) (27%).

The study also revealed that 66% of respondents overall said that they are very interested or moderately interested in the concept of unified fabric or SAN/LAN convergence.

The reason for the iSCSI interest is mostly due to the adoption of server virtualization technology. Forrester claims that iSCSI is growing as protocol of choice for virtual server deployments. In a Forrester Research survey from January 2009, Fibre Channel led strongly in protocol selection for virtual server connection. This time around, Fibre Channel still leads the way, but iSCSI is closer behind.

According to Forrester, the data points to strong interest in unified fabrics, as well as interest in iSCSI in 10GbE format as a viable alternative to Fibre Channel storage.

Travis Vigil, a senior manager with Dell responsible for the company's line of EqualLogic iSCSI SAN products, says iSCSI was one of the few segments of the storage market to experience year-over-year growth of approximately 25%.

"The economic situation was an accelerator for iSCSI growth as customers looked to eliminate cost and complexity from their environments. The other contributing factor to the growth of iSCSI is server virtualization as we've seen more and more customers choose iSCSI as their fabric of choice," he says. "First time server virtualization customers find iSCSI to be a cost-effective technology."

Vigil also says iSCSI 10GbE will only make iSCSI better as Ethernet becomes the basis for unified network fabrics. "Fibre Channel over Ethernet is still developing, but iSCSI is here and now. Ten Gigabit Ethernet is only going to make iSCSI better," he says.

For users with legacy Fibre Channel gear, FCoE leads the way as the Ethernet choice. If they move to an Ethernet SAN, 44% chose FCoE, but 30% chose iSCSI, at a close second, and 26% chose NFS, in third.

The shift to unified fabrics is definitely underway, but it will be interesting to see how iSCSI fares as FCoE products become fully baked.

Friday, January 22, 2010

IBM, FujiFilm demo 35TB tape cartridge

January 22, 2010 -- As the Linear Tape-Open (LTO) Program announces licensing details for the next generation of 3TB LTO 5 tapes, IBM and FujiFilm are unveiling new technology that makes it possible to hold up to 35TB of uncompressed data on a single tape cartridge.

The world record breakthrough was made possible by an improvement in the precision of controlling the position of the read-write heads, according to IBM. The pinpoint control yields better than a 25-fold increase in the number of tracks that can be squeezed onto the half-inch-wide tape.

The scientists have also developed new detection methods to increase the accuracy of reading the tiny magnetic bits, an advance that increases the linear recording density by more than 50%.

The tape also uses a new, low-friction read-write head developed by IBM Research.

IBM claims the demonstration (view the IBM Research video) was performed at product-level tape speeds (2 meters per second) and achieved error rates that are correctable using standard error-correction techniques to meet IBM's performance specification for its LTO Generation 4 products.

Tape still has the advantage over hard disk drives (HDDs) when it comes to cost. IBM claims today's tape systems cost one-fifth to one-tenth the price of disk-based storage systems, not to mention the power savings associated with magnetic storage.

The concept of storing that much data on a single tape may stave off the "tape is dead" argument for another decade – at least that what IBM is hoping.

Wednesday, January 13, 2010

Survey: DR plans slipping through the cracks

January 13, 2010 -- I happened to be on the phone with Symantec last week to discuss the results of their annual State of the Data Center Study when an earthquake shook California. In my mind, the event reiterated the need for frequent testing of disaster recovery (DR) plans. That's why I was surprised to learn that a growing number of users are letting DR plans go untouched for long periods of time.

Once the quake subsided, Matthew Lodge, a senior director with Symantec's information management group, told me that DR plans are vulnerable and have become a victim of a decrease in funding.

According to the Symantec 2010 State of the Data Center Study, which is based on surveys of 1,780 data center managers in 26 countries in November 2009, there is room for improvement in disaster recovery.

One-third of those surveyed said their DR plans are undocumented or need work and important IT components, such as cloud computing, remote office and virtual servers are often not included in the DR plan. To make matters worse, almost 33% of enterprises haven't re-evaluated their disaster recovery plan in the last 12 months.

"One of the hardest hit areas of the data center in terms of funding is disaster recovery," said Lodge. "A lot of companies haven't refreshed or reevaluated their DR plan in quite some time."

It seems IT staffers are focusing their efforts elsewhere. The survey revealed the top concerns in the data to be increased complexity and too many applications.

One-third of all enterprises say staff productivity is hampered by too many applications. Adding to the complexity is the continued increase in data causing 71% of organizations to consider data reduction technologies such as deduplication, according to Symantec.

Most enterprises have 10 or more data center initiatives rated as somewhat or absolutely important and 50% expect "significant" changes to their data centers in 2010.

Lodge said 50% of all enterprises say applications are growing somewhat/quickly and half are finding it difficult and costly to meet service level agreements (SLAs).

Mid-sized enterprise data centers are weathering the storm best. Lodge said mid-sized enterprises are more agile, show more activity, and predict major changes to the data center and new applications in 2010. Mid-sized enterprises also place a higher importance on staffing and training than their small or large enterprise counterparts.

Mid-sized enterprises are more aggressive and pioneering than either small or large enterprises. They are adopting new technology initiatives such as cloud computing, replication, and deduplication at 11-17 percent higher rates than small or large enterprises.

Not surprisingly, Symantec's answer to the aforementioned challenges is software. "Let the software help you out. Users need to look for more areas of integration and need to adopt an automated, policy-based approach to management rather than going for a raft of individual tools for the data center," Lodge said.

Some additional tidbits from the Symantec 2010 State of the Data Center Study:

Security, backup and recovery, and continuous data protection are the most important initiatives in 2010, ahead of virtualization.

Staffing and budgets remain tight with half of all enterprises reporting they are somewhat/extremely understaffed.

Virtual machine protection continues to be a focus for enterprises, with 82% of enterprises considering virtual-machine technologies in 2010

Check out the InfoStor disaster recovery page for the latest news and view on the state of DR.