The Pinnacle Blog

Latest news, industry updates, helpful business advice, and ways technology can improve your business. The Pinnacle Blog

EricStorage Without Bloating Your IT Budget

Filed under: IT/Networking | Technology Planning

For the business leader:

 Has your IT department asked for a large storage budget this year leaving you concerned with the rising costs of storage? You can now purchase your IT storage as you need it in small discreet chunks using new storage technologies rather than in large expensive chunks. 

There are currently two companies that stand out in this market.  HP became one of the strongest players with the purchase of Left-Hand Systems.  The second company is a new player in the market originating in Indiana, Scale ComputingScale Computing leverages the same technology as HP with a promise of much less expensive price.

What do I need to know about storage and why should I care?

  • You can purchase storage in small increments as your business grows rather than large investments based on best guess projections.
  • Usable clustered storage runs from $5-10 per GB, while traditional SAN storage runs $12-24 per GB after configuration and software licensing.
  • The potential ROI in clustered storage is the amount of dollars/yr investment vs. available storage at year end. Reducing available, but unused storage purchases, eliminating “cage” costs, and eliminating software licensing can dramatically increase storage ROI’s.
  • If you currently have a terabyte of storage, are planning a virtualization project, or getting ready to renew maintenance on an existing SAN - you should consider the benefits of clustered storage and discuss with your IT Team or provider.
  • If you’ve outgrown your current storage solution and are planning to expand this year, a cheaper alternative may be clustered storage, which can be less than maintenance on existing systems
  • Like all technology considerations, your storage solutions should be part of your overall Technology Plan.

 

For the Technician:

 JBOD to JBODS
SAN technologies first took off when innovative engineers discovered they could leverage the low costs of hard-drives built for Personal Computers to build large virtual storage systems. Enterprising engineers found that redundant arrays of inexpensive drives allowed them to build large pools of storage leveraging consumer focused drive technologies. Engineers soon learned that software could also take advantage of the larger number of read devices to cheaply produce inexpensive, yet fast storage systems. JBOD (for Just a Bunch of Drives) philosophies for building storage solutions eventually became so commonplace they overtook the strategies for building large storage systems. JBOD designs now dominate all multi-drive storage solutions.

Scale Computing and HP’s Left-Hand Systems take advantage of increasingly inexpensive multi-core CPU based server hardware in much the same way engineers first took advantage of drives with increasingly larger number of read/write mechanisms. Both vendors’ architectures replace the concept of “cages of drives” with the concept of storage nodes.   Storage nodes are essentially 1U servers filled to the brim with drives. Each storage node is a 1U server with four internal hard drives configured for redundancy. SANs are built by plugging multiple storage nodes into a common GB Ethernet switch. Think of the new storage approach as JBODS (just a bunch of drives and servers) instead of JBOD.
The advantages of the clustered storage node approach to building SANs over cage architectures are many.

  • Eliminating traditional cage frameworks removes cage capacity issues, as both vendor architectures scale to petabytes from even a small 3TB starter SAN.
  • Adding storage nodes also increases the number of available read/write mechanisms, and therefore SAN performance; while at the same time increasing storage access bandwidth as single, or sometimes multiple, GB uplinks are added with each new node.
  • Storage nodes can be added “as needed”, in small or large increments, with both architectures allowing nodes consisting of different sized drives to be added to the SAN.

Scale Computing and HP’s Left-Hand Systems architectures leverage software to ensure data is replicated across different storage nodes, ensuring entire storage nodes or drives can fail without losing data. While Scale and Left-Hand architectures differ in approach to providing drive-management redundancies, both offer increased redundancies potentially dwarfing the majority of caged SAN solutions. Both vendors promise some version of “all-software, all-the-time” with their SAN purchases; and both vendors indicate their solutions include previously enterprise class storage management solutions at no additional cost.  

Sizing and Buy-in Costs
Both Scale Computing and Left-Hand Systems price storage nodes at usable sizes (size after configuration for redundancy). Today’s nodes start at 1TB each, with three nodes required to start a SAN. Pricing appears to vary broadly depending on vendor, hardware, and options. Pricing starts at under $13K for 3TB of Scale branded blades; Scale also offers IBM branded blades at higher premiums. HP branded Left-Hand pricing starts in the $25 -30K range for a similar amount of storage. High availability options with power and network options are available at higher pricing from both vendors.

Conclusion
Whether an ROI driven business executive, or a technology driven gear-head – both of these technologies are worth watching and perhaps even testing. The cost advantages inherent in JBODS design closely mirrors past JBOD philosophies which have shaped the storage industry for years. Combining IP based architectures with commodity based clustered CPU and drive technologies, at a time when thousands of terabytes are referred to as petabytes, is somewhat akin to launching a big-wave surfer past giant 40-foot breakers with a jet sky; whether they crash and burn, or come out on top of the surf in the end – clustered storage solutions are guaranteed for a very, very wild ride.

Comments

Add comment




biuquote
  • Comment
  • Preview
Loading



Log in