Brian Garrett's Most Recent Posts

Cisco announced a series of upgrades to their MDS line of storage switches at Storage Networking World in San Diego last week.  Maximum port counts were upgraded to a whopping 528 ports per director; 4 and 10 Gbps support was added across the entire MDS product lineup; and a new larger director chassis was introduced.

In my opinion the bottom line with this announcement is two-fold:

1. Cisco has the biggest director on the planet from a port count perspective

Cisco now tops out at 528 ports per director compared to 256 ports for McData and Brocade.   Brocade and McData argue that you can't use all of those ports at full speed at the same time which is true - only 132 ports can be run simultaneously at 4 Gbps.  I contend that this "over-subscription"  issue is a non-starter for most data center managers.   Fibre Channel networks are typically configured in an over-subscribed manner. Each port into a shared storage system is typically shared by many servers - up to eight or more is common.  Yeah, full bandwidth into and out of all ports in a director is architecturally more predictable, but given the low utilization rates on most FC networks, and the trends towards scaled-out clustered architectures and server consolidation, I believe Cisco with more ports per director is solving the more pressing need  - more connectivity.  In other words, I believe that for the majority of users, more ports in a singly managed platform is more desirable than guaranteed line rate on every port.

The most likely component to fail in a computer system is a hard disk drive. Like your car, the mechanical parts are more likely to fail than the pure electrical parts.   For example, a wheel bearing in your car is more likely to fail than the computer chip that monitors your engine's performance.  Similarly, a wheel bearing in a hard drive is more likely to break than the processor on a disk controller.  The Mean Time Between Failure (or MTBF) is a rating used in the computer industry that attempts to predict the likelihood of a failure.  More specifically, the MTBF is a measure of the predicted time between failures.

I was reading the March 13, 2006 issue of Business Week and stumbled across a fascinating article by Brian Grow on the emerging field of quantum cryptography.  Brian describes how particles of light, or photons, are  being used to intermingle "decoy" particles of light into an optical  data transmission stream using a private key known only to the sender and receiver.  Such transmissions aren't tamper-proof, but under the laws of physics, any attempt by hackers to inspect or insert photons triggers can be noticed.  This fascinating science published in the Feb 24th issue of Physical Review Letters got me to thinking about the steps that end users should be considering to secure corporate data assets.

Hu Yoshita, the CTO of Hitachi Data Systems, reacted to my blog entry on the Storage Controller of the Future in a recent blog entry. I suggested that the enterprise-class storage controller of the future would look more like one of the emerging commodity-based clustered storage controllers (e.g. Equallogic, Exanet, Isilon, Lefthand Networks) than the currently shipping enterprise-class solutions from EMC, HDS and IBM (e.g. the HDS TagmaStore USP). Hu suggested that while emerging clustered controller architectures seem appropriate for tier 2/3 storage needs, mission-critical enterprise-class applications are ill-suited for these platforms. I believe that depending on the time frame we are talking about, and the way that the architecture is characterized, we both may be right.

Last week Qlogic announced the acquisition of an Infiniband silicon provider named Pathscale, Inc. for $109M. Although Qlogic is the latest storage-focused company to purchase Infiniband technology, it is not the first company in the storage market to get on the Infiniband-wagon. Cisco acquired an Infiniband company named Topspin for $250M last year. Engenio, SGI and Data Direct recently began shipping storage controller with Infiniband support for connecting to servers. Storage controllers from vendors like Netapp, HDS and Isilon are using Infiniband inside as a backbone or cluster interconnect. So why have vendors in the storage market started jumping onto the Infiniband-wagon?

As an analyst at ESG I've seen a bunch of storage controller designs from startups over the past couple years. Most have traits in common that are very unlike the enterprise-class storage controllers that dominate the high end of the market today. Emerging storage controllers are being built using commodity components. Intel servers clustered together into a singly managed scalable system is the rule. Most are using commodity SATA drives -- the same drives being used in entry level servers. Most are built using Linux instead of a homegrown or embedded operating system.

So what does the enterprise-class storage controller of the future look like? I believe it will look more like an emerging clustered based storage controller than the currently shipping enterprise-class solutions from EMC, HDS and IBM. Today clustered storage controllers communicate with each other using Gigabit Ethernet or Fiber Channel. In the future I believe we'll see 10GigE, or Infiniband, or ASI for PCI Express. Today the drives are 3.5" . In the future we'll see 2.5" and even 1" drives. Today clustered storage controllers use off-the shelf adapters to connect to servers and hard drives. In the future we'll see adapters with programmable offload, compression, encryption and replication built-in. Gone will be the proprietary bus architectures of legacy enterprise-class storage controllers. Gone will be the custom hardware connected to that proprietary bus. Gone will be the cost of all that custom hardware. Over time I believe that enterprise-class storage controllers built using standardized commodity components will be the exception rather than the rule as it is today.

The "Storage Products of the Year" in the February, 2006 issue of Storage Magazine has me thinking about the future of backup and archiving. The backup hardware awards went to a pair of tape libraries and a compressing disk based backup appliance. The tape library systems were recognized for support of four different media types (Sun L1400M) and for modular field upgradeability (Quantum PX500). The disk based appliance was recognized for compression rates of twenty to one and compatibility with existing backup software (Data Domain DD460). Taken together, I feel these products point towards the future of long term archiving – why not build it all into a single box? I think that the ultimate backup and archiving solution of the future would bring out the best of each of these technologies:

ESG estimates that 75% of the data on a corporate network resides outside the data center. How are all of those remote servers being backed up? And what about the remote desktops and laptops? Besides the risk associated with lost productivity and lost work in remote offices, what about the risk at the corporate level of a law suit or a fine for a regulatory mandated regulation?

Tacit acquired a software company called Mobiliti this week. Mobiliti software makes remote shared drives appear local when users are disconnected from the WAN and provides backup service for desktops and laptops over the WAN. Tacit is a startup that began in the wide area file service (WAFS) market and is now branching out to provide application acceleration and protection services over the WAN. This acquisition personifies consolidation trends happening in the data processing industry over the past couple years:

• IT administrators are consolidating servers, storage and services in the data center
• IT administrators want to extend the value of consolidation out to the edge of the network into remote offices and for mobile users where trained IT staff is in short supply
• IT administrators want to consolidate remote application(e.g. Email), storage (e.g. shared files and backup) and networking (e.g. DNS, firewall) and print services.
• Storage, networking and systems vendors have recognized the value of the consolidated remote services and a feeding frenzy of M&A and consolidation activity has begun.

Delivering applications to remote and branch offices over a wide area network is slow, expensive and complex. The Enterprise Strategy Group (ESG) estimates that as much as 75% of the data on a corporate network resides outside the corporate data center, where it is often unprotected and where IT staff is in short supply.

Vendors in the networking, storage, systems and software markets have all realized that fixing the remote IT problem is a huge opportunity.   Regardless of the name of the technology (WAFS, WAAS, WADS, WAN acceleration, WAN optimization,..), the stakes are high and the market is hot - more than $1B in revenue, double digit growth and close to $1B in M&A activity over the last two years:  Cisco acquired Actona ($82M ) and FineGround ($78M), Juniper acquired Perabit ($337M) and Redline($132M), Citrix acquired Netscaler ($300M) and Brocade made a strategic investment in Tacit($7.5M).  Systems, software, storage and networking companies including HP, IBM, Microsoft and Network Appliance have aligned with startups and are bringing products to market.  Remote backup is an adjacent market that is also red hot.  Companies providing accelerated and centralized backup and recovery for remote users include Asigra, Avamar, Signiant and FilesX.

There is a long list of startups that have not yet been acquired in this category - and a similarly long list of potential acquirers. Like a game of musical chairs, the list of potential acquirers is longer than the list of companies that can be acquired.  I’m not sure who’s going to be left standing, but it seems clear that bidding wars are likely in 2006.  My picks for the most likely to be bought in the near future are Tacit, Riverbed and Packateer.