TechRepublic's free Storage NetNote newsletter is designed to help you manage the critical data in your enterprise. Automatically sign up today!
Previously in this series about choosing an iSCSI storage solution, I told you that we had whittled down the iSCSI SAN vendor selection process to two prospects: LeftHand Networks and EqualLogic. The consulting firm that we use for major projects had just partnered with LeftHand, and it was while researching its solution that I actually stumbled across the EqualLogic product.
Both solutions offered a wealth of included features, including the ability to perform volume snapshots as well as to replicate the unit's data to a redundant device in a disaster recovery location. Both units also boasted extremely good management software that (*gasp*) made sense.
Comparing the LeftHand and EqualLogic storage solutions
I settled on the LeftHand and EqualLogic systems for a variety of reasons:
- For both systems, the management is top notch and extremely easy to use.
- Both solutions are fully redundant and neither has a single point of failure, which is a critical component for our storage infrastructure.
- LeftHand and EqualLogic are both iSCSI-only companies, so they have more of a stake in making sure their iSCSI offerings work perfectly and competitively against the bigger players.
- Both companies are financially strong, so they should be around a while.
- References from companies that use both products came back as excellent in every case.
- Everything I needed and wanted is included in the purchase price with no add-ons that drive up the cost.
From a purely technical perspective, the LeftHand Networks NSM (Network Storage Module) units and the EqualLogic PS (PeerStorage) arrays are fairly on par with each other. Both solutions offer redundancy, seemingly good performance, and the ability to expand to well beyond 100 TB without the need to take down the array. In fact, both solutions work on the concept of clustering. If you need more storage capacity, buy it; add the new unit to the cluster; and watch as your data is automatically striped across the new hardware.
With both units, performance increases as the number of individual array units increases. Each unit has its own network connections and processing power. Since data can be striped across multiple units for redundancy, the additional network connections increase the total bandwidth available to the array cluster, and the additional processors in new units provide more horsepower for the array cluster as a whole.
What are the differences?
LeftHand and EqualLogic have very different ideas about how to achieve total solution availability and reliability. Whereas LeftHand prefers to achieve redundancy by installing three separate hardware devices, EqualLogic achieves the same goal in a single device in which every piece of hardware is fully redundant. As such, LeftHand's NSM units have less capacity per device than EqualLogic. In a single unit, LeftHand tops out at 2 TB while EqualLogic's PS200 maxes out at 5.6 TB.
LeftHand NSMs have two gigabit-Ethernet jacks while each EqualLogic unit has six, of which only three are active. The other three are present on a backup controller to provide redundancy. In looking at a single EqualLogic array with 5.6 TB (which would be roughly three LeftHand NSM 200s each with 2 TB), it's easy to tell that the LeftHand solution, with two available, active jacks on each unit, provides more total bandwidth to the array.
Of course, performance is also an important aspect of the selected array. To be honest, this area was extremely difficult to quantify. EqualLogic quoted me insanely high numbers (~50,000 operations per second) while LeftHand's number came in insanely low (~300 - ~1,200 operations per second per array). EqualLogic derived its numbers by reading directly from cache, whereas LeftHand uses more real-world numbers. I suspect that the truth lies somewhere in the middle.
In the end, we chose a 5.-TB PS200E array from EqualLogic, based on existing customer feedback, multiple vendor demonstrations, superior management software, and a 25 percent price differential in the per-TB cost of the solution. Personally, I also prefer the "single box", "fully redundant" scenario afforded by the PS200E.
In my next article, I'll detail the installation of the array, which is where I realized that we definitely made a good decision.