Why Nimble Storage Is Growing So Rapidly - A View From The Trenches

| About: Nimble Storage, (NMBL)
This article is now exclusive for PRO subscribers.

(For brevity, I will use NMBL to refer to both the stock and the company.)

The recent article by Technology Assessment Group got me thinking.

-----------------------

That article mentioned that the success of NMBL has validated the customer demand for hybrid arrays. It also mentioned that for the following reasons, NMBL will find it difficult to gain acceptance in Tier 1 Enterprise and Data Center markets:

1. Active/Standby controllers

2. Provision of only one protocol: iSCSI

3. Lack of deduplication

-----------------------

I manage network storage arrays for a living. I have direct, hands-on experience working with many arrays, including ones from Nimble Storage. This article is a view from the trenches.

NMBL has been described as one with a "Disruptive Technology", and a "Game Changer". I would be inclined to agree.

NMBL makes hybrid arrays, but not all hybrid arrays are the same. For most arrays, performance is determined by the number of FC/SAS/SATA drives behind the controllers. Expensive SSD drives are typically used for caching. The highest performance arrays are all SSD, and as a result, several times more expensive compared to the hybrid arrays of the same capacity.

A quick look at the technology in NMBL's arrays:

In a Nimble Storage array, data is first written to RAM. It is then compressed, and then written sequentially, in variable length blocks, to low-cost SATA hard disks. SSDs are used for read caching.

Using low-cost hardware, Nimble arrays are able to achieve sub-millisecond latency, which in the real world, we only see in all SSD arrays. For those who have experienced "click-and-wait", latency matters.

Using compression and high-capacity SATA drives, NMBL is able to provide a 50TB array in a 5.25 inch-high unit. Compare that with a 50TB all-SAS array from EMC/Netapp/HDS/HP-3PAR. The SAS array would take an entire rack of space, but would still not be able to match the performance of the NMBL array.

NMBL also bundles in backup and replication software.

In a data center, space usage matters:

I personally know several storage administrators who have replaced 3 full racks of storage and backup equipment with less than half a rack of NMBL arrays.

Simple economics: Each rack in a data center costs about $5K per month. That comes to $60K per year, or $180K over 3 years. In 3 years, most companies refresh their equipment. The space saving of just one rack in a data center pays for the purchase of the half rack of NMBL arrays.

So why is NMBL growing so rapidly? SMB customers of NMBL realize that they are able to get a high-performance array that saves data center space and power, is way easier to manage, has a better ROI, and has support that is second to none.

How about reliability and the need to support a 24x7 operation? This is a core requirement for all users, not just for Tier 1 Enterprises.

Most arrays, including NMBL, are designed to have no single point of hardware failure. Hardware fails. As long as the redundant component takes over, business operations are not impacted. No harm done. The defective component is, in most cases, swapped out without needing a downtime.

So does that mean all arrays are suitable for 24x7 operation? Not quite. The devil is in the details.

Once or twice a year, there is a need install patches and do software upgrades. For many arrays, it means a reboot. For an IT team, it means scheduling a time, usually a weekend night, when business operations will come to a halt. Critical business applications and databases will be shut down before the reboot; then brought up, and tested after the reboot. One simple software upgrade typically take over a 100 man hours of effort.

With a NMBL array, it's a few clicks in the web interface, done in the middle of the business day. The array installs the software upgrade on the passive controller, and seamlessly switches to it from the active controller. No application downtime. No weekend nights spent in a data center.

If you have a requirement that for 24x7 operations you would like to only consider arrays that can do software upgrades without downtime, then you have very few choices. NMBL is one of them.

------------------------------

The following don't really look like gating factors to me:

1. Active/Standby controllers:

Compared to the cost of the array, the hardware cost of a controller is miniscule. I prefer the NMBL design of keeping the standby controller passive till it is needed.

2. Provision of only one protocol: iSCSI

That certainly limits both the use cases and the market for the NMBL array. But as clearly evidenced by the rapid growth in sales, NMBL is still a long way from reaching any limits in the market for iSCSI storage.

3. Lack of deduplication

As a customer, I just look at how much data I can store per $. Deduplication is a good option for storage of archival data. If deduplication is turned on, performance takes a big hit. That makes deduplication unsuitable for primary storage.

--------------------------

There is a lot more that NMBL can grow, particularly if they decide to have solutions for the FC and CIFS/NFS file services markets.

Competition: Some competitors seem to have bits and pieces of the technologies that make NMBL successful. Virsto (now VMware) does write caching in a way similar to NMBL. FIO, Sandisk, EMC, and several others use SSDs for read caching. However, none seem to have really put it all together. To complete with NMBL's technology advantage, the big players, EMC/Netapp/HDS/IBM, might not only have to redesign their products but might also end up cannibalizing their own sales of traditional SAS arrays.

Why is NMBL successful in the SMB market but not so much in Tier 1 Enterprises? For one, in the SMB market, the decision-making process is a lot simpler and quicker. Enterprises tend to place a greater emphasis on standardization, maintaining relationships with existing vendors, and having defensible purchases (as in - no one ever got fired for buying from IBM). As NMBL becomes more established, I can see it getting a higher share of the Tier 1 Enterprise market.

Your mileage may vary, but in my opinion, NMBL stock is a buy.

Disclosure: I am long NMBL. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Additional disclosure: I am a network storage administrator for a Tier-1 enterprise.