Organization of Content
What is a Hybrid Storage Array and why would anybody want to buy it?
Addressable Market and Market Segmentation
Discussion of Gating Factors
High Availability Architecture
Storage Management Software Support
Brief discussion of emergent vendors that will compete with Nimble
A - Tier 1 Enterprise Storage Requirements
B - Storage Use Cases
Nimble Storage, Inc. (NYSE:NMBL) has validated customer demand for Hybrid Storage Arrays, which are built from both flash memory and traditional hard disk drives. Nimble stock is back near its high of $49.00 and is about 130% of its December 13, 2013 IPO price of $21.00. New customer acquisition rate is strong. Clearly, Nimble Storage IPO has been successful in building a product for the SMB (Small Medium Business) market.
Key questions are: How far can this stock go? What are the gating factors?
These questions are relevant, given the failure of the Violin Memory (NYSE:VMEM) IPO, September 27, 2013
The purpose of this article is to help investors build their own financial models. To this end, this article characterizes upside potential and the competitive environment.
These results are consistent with my article published in Seeking Alpha October 4, 2013.
Nimble Storage unit sales are improving briskly, while operating losses have lessened to 99% of total revenue. (99% figure from Don Dion). On January 15, 2014, Nimble stock price recovered 16% in one day. Year over year, customer count increased from 780 to 2121 for the Q3FY13 to Q3FY14 time period. Gross margins improved from 62.3% to 64.5%. Total revenue, YOY, improved 284% from FY13 $14 M to FY14 $53 M. Momentum is in the correct direction
With respect to how far the stock can go, I recommend caution since the stock has already run 130% over its IPO price.
The caution is more about avoiding "Euphoria" than having any major negative views on the company or product. There are a few lessons to be learned from the Violin IPO, and one of them is not to get too caught up in the " Violin flash is fast and therefore EMC will be driven out of business" thought process. Clearly, Nimble products are accepted in the market place, but there are gating factors to revenue, margins, and the Nimble stock price.
Nimble has three key gating factors. The first is that their Active/Standby architecture is more suitable for the SMB (Small Medium Business) market place. At the departmental and divisional level of large corporations, the corporate IT department provides more rigorous guidance, and an Active/Active HA architecture is likely to be preferred. In addition, at the corporate divisional and departmental level, there may be a proliferation of workloads, and wider variety of protocols, beyond iSCSI, will be needed. Finally, lack of deduplication, in context of unstructured data growing at 60% per year, may create and opening for other emerging vendors to compete.
There are other Hybrid and All Flash Vendors in mid-range space, and they are credible competitors. These vendors include but are not limited to general purpose vendors such as Compellent/Dell, Tegile, Pure Storage, and niche vendors such Tintri and Nutanix.
According to IDC and Gartner, the mid-range storage market is roughly in the $18 B range and growing about 12% per year on its way to becoming nearly 50% of the entire storage market.
While Nimble is clearly more successful than the September 27, 2013 Violin Memory IPO which saw a roughly 70% decline from its IPO initial prices, there are limitations to the upside. Nimble (along with its competitors) will remain in the mid-range market, and not break into Tier 1 Enterprise Market.
Nimble not entering the Tier 1 Enterprise market is not a serious issue, since the mid-range market is large enough to support many vendors. However, there are emergent vendors that will compete with Nimble in the mid-range space, and limit upside revenue gains. Key competitors include but are not limited to general purpose vendors such Dell/Compellent, Tegile, Pure Storage, and niche vendors such as Tintri and Nutanix.
What is a Hybrid Storage Array and why would customers want to buy it?
Nimble Storage makes a product called a Hybrid Storage Array. It consists of flash memory and hard disk drives, as well as storage management software. Flash memory provides superior performance, occupies less space while requiring less power and cooling. However, flash memory is quite expensive relative to disk. For example, 1 GB of commodity flash memory is priced about $1 per GB. In contrast, hard disk drives provide superior capacity and are priced about 4 cents per GB but do not provide the performance of flash. The benefit of a Hybrid Storage Array is that it provides performance and capacity, at a reasonable price to a customer, and is a very practical storage solution.
This is in contrast to an All Flash Array, such as provided by Pure Storage, Violin, and Skyera. These products are optimized for performance; these products are clearly useful in applications with smaller amounts of data that need a lot of performance. A couple of common applications are report generation and billing. Of course, there are all sort of marketing claims that All Flash Arrays are just as cost effective as disk, and these arguments focus on data reduction features such as deduplication, compression, and thin provisioning. What they forget to tell you is there is nothing new about these features and many (but not all) vendors provide the same data reduction technologies that can also be applied to hard disk drives. Or there will claims of 10X deduplication ratios which are not consistently achievable. For Tier 1 Enterprise Storage usage, quantities of data are spoken of in the Petabyte, and Exabyte range and the superior economics (with acceptable lower performance) suggest that there is still a very useful role for hard disk drives for less frequently accessed data. The huge amounts of data growth discussed in the media are centered on unstructured data that has a growth rate of about 60% per year. Examples of unstructured data include Word, Excel, Power Points, photos, videos, and audio files; the key point is that these kinds of data consume a lot of space. For example, a photo often consumes 2 MB, and a large spreadsheet or power point can consume 10 MB to 20 MB. In contrast, a single bank deposit transaction used in Retail Bank OLTP (Online Transaction Processing) can consume only 50 bytes (think 50 type written characters).
In summary, All Flash Arrays are useful when performance is the top priority, but are more practical with smaller quantities of data. On the other hand, Hybrid Storage Arrays provide a balanced combination of performance, capacity, and cost effectiveness.
The key take away is that there is clear demand for Hybrid Storage Arrays, and the combined benefits of performance, capacity, and cost-effectiveness have been validated by the success of the Nimble IPO and customer acquisition numbers.
Addressable Market and Market Segmentation
To understand the source of revenue for Nimble, I have created a market segmentation chart, as per Figure 1 below.
There are two key take-aways:
1. Nimble will be deriving its revenue from the Midrange, Commercial market segment as depicted in Row 3 below. But will have several competitors which include but are not limited to Compellent/Dell,Tegile, Pure, Tintri, and Nutanix.
2. Nimble will not be deriving any significant revenue from market segments Rows 2, 4, 5, and 6.
Discussion of Gating Factors
From a product perspective, three limitations will serve as gating factors:
1. Active/Standby Controllers
2. Provision of only one protocol: iSCSI
3. Lack of deduplication
For customers, who do not need these functionalities, Nimble will continue to be an acceptable product.
For those customers who do need additional functionality, other vendors will be legitimate competitors to Nimble
Nimble's Active/Standbay controller architecture will limit their ability to break out into higher end customers. Nimble's reliance only on the iSCSI protocol will exclude them from competing for customers who wish to use CIFS, NFS, and FC protocols. In addition, Nimble does not provide deduplication. Lack of these features should not be interpreted as suggesting that Nimble is a poorly designed machine. Clearly, customers are buying Nimble product, and are making repeat purchases, and by inference are reasonably happy. The recommended way to interpret these limitations is that Nimble product is experiencing success at the SMB subset of the mid-range market segment. But Nimble will experience challenges with more sophisticated customers at the corporate departmental and divisional level, where more rigorous advice is dispensed by corporate IT, and they will likely seek an Active/Active HA architecture. For a small company, Nimble did a good job bringing a reasonable feature set to the market place. But Nimble's limitations leave openings for other companies to compete directly with Nimble by addressing requirements that Nimble does not fulfill.
Please refer to technology section below for more details.
The reason why upside gating factors are highlighted is that I have observed both professional and non-professional investors focus on a set of success factors which is too small, forecast huge revenue upside, and find themselves hugely disappointed. Violin Memory is a prime example of a stock that disappointed in a spectacular manner. Some examples of the thinking which occurred around Violin Memory included (paraphrased composite comments for brevity) "They're the fastest" and "the best technology" and will therefore put EMC out of business while penetrating the Fortune 100. I view this level of analysis to be substantially insufficient with respect to deploying corporate-wide Tier 1 Enterprise applications into the Fortune 100. This level of analysis also provides little insight into a vendor's ability to consistently win customer sales engagements against credible competitors. This level of analysis seems to miss the point that flash memory is no longer new, and there are many competent vendors engaged in this market space. To put this in context, I saw my first flash based prototype system late 2006, and have deployed several flash based systems in since 2008. That's nearly 8 years of history with this technology. It takes much for than just being fast to win in this space.
Clearly, none of vendors mid-range vendor listed this article have all the functionality to server as Tier 1 Enterprise Storage.
See Appendix A detailing the requirements in Tier 1 Enterprise Storage
EMC, HDS, and IBM flagships products (VMAX, VSP, and DS-8000) respectively do not compete with these mid-range vendors. These legacy products will be replaced by the flash-capable next generation as provided, respectively, by EMC, HDS, and IBM.
This is a gating factor on the upside revenue potential for Nimble (as well as the other storage mid-range vendors).
But these mid-range vendors, which include but are not limited to Nimble, Compellent/Dell, Tegile, Pure, Tintri, and Nutanix, do provide sufficient functionality for the mid-range, in a manner far less costly than the EMC, HDS, IBM flagship products, and are much simpler to install and administer. So they will all do a good Job in the mid-range and will compete in their respective areas of advantage. In fact, the Tier 1 Enterprise market growth is de-accelerating while requiring heavy investments and so being in the mid-range is a good place to be.
But with respect to Nimble, the key take away is "expect growth in revenues, but Nimble will have competitors which will be another gating factor on revenue growth." Nimble, Compellent/Dell, Tegile, Pure, , and Nutanix and Tintri will all get a piece of a nice sized pie that is growing. Nimble will not be the single vendor to deliver a "knock out" blow to the other vendors. It's a highly competitive free market at work.
High Availability Architecture
Currently, Nimble Storage uses Active/Standby controllers. This is a very fundamentally different from their competitors, who use Active/Active controllers. This means that each array has two controllers in case one fails. But in an Active/Active HA architecture, both controllers are processing data. In an Active/Standby HA architecture, only one controller is processing data, and the other is idle. So if the controller A fails, there an out-of-service condition while the failover operation occurs to bring the idle controller B into production. In an Active/Active HA controller architecture, both controllers are processing data. If one controller fails, the remaining controller continues to maintain service to the end user. The end users on the failed controller are then "failed over" to the second controller. In a very good HA failover architecture, the failover occurs very rapidly, and the end user may notice a very short pause and slower response time. These conditions are considered to be acceptable, since the alternative it to be completely out of service. Another benefit of the Active/Active HA architecture is better ROI (Return on Investment). With Active/Active HA architecture, both controllers are processing data, which means more work is being done for the money invested in equipment. In an Active/Standby, where the second controller is idle, the customer is paying for equipment which is idle, which negatively impacts ROI.
I was surprised to learn that that Nimble had an Active/Standby Architecture; I expected a product that has been around this long to be Active/Active because most serious storage companies deploy an active/active HA architecture. Sophisticated customers (such as those at the corporate departmental and divisional level) who understand these architectures are more likely to buy Active/Active HA architecture (and have the money to do so). Less sophisticated SMB customers who may not understand this and don't have the money, are more likely to buy the Active/Standby Architecture. This may be a significant point as other mid-range storage vendors (such as Tegile and Pure who have Active/Active HA architectures) gain more traction in the market place.
Hybrid and All Flash Arrays
While Nimble has clearly validated the market for Hybrid Storage Arrays, a number of other vendors are selling into this space. They vary from general purpose (such as Tegile and Compellent/Dell), Flash Only (NASDAQ:PURE) and niche (Tintri and Nutanix). But all of these vendors do address specific customer requirements in the mid-range storage space.
The next limitation is protocol support. Nimble Storage supports iSCSI only.
The legacy vendors such as EMC VNX and NetApp tend to have very broad protocol support, encompassing iSCSI, CIFS, NFS, and FC, and often refer to this category as "Unified Storage". Unified Storage allows a customer to deploy one set of infrastructure to serve multiple workloads (as represented by different protocols) and leverage a common set of people, processes, and IT skills all accessing their infrastructure through a common interface. Clearly, the mid-range vendors provide a limited offering in this space, relative to the legacy vendors, with the exception of Tegile, with its full set of protocols.The significance of this is that CIFS is typically supported on Microsoft Windows products, and they are clearly a large part of the target market. NFS is widely used by sophisticated customers in higher end file sharing. In addition, VMware is increasing NFS support because it is easier to deploy and manage. FC is important for corporate shops running mission critical databases because of the manageability and deterministic response. Nimble, being only iSCSI, may have to by-pass some of these revenue opportunities.
Mid-Range Storage Management Software Stack (limited view for brevity)
Brief discussion of emergent vendors that will compete with NimbleA couple of comments regarding Deduplication. It a very popular tool to use in virtualized environments such as VMware to eliminate redundant system images, such as what is stored in VMware VMDK files. Both Nimble and Compellent/Dell do not have deduplication, and therefore are unable to remove duplicate VMDK files. Nimble does have compression, which reduces the amount of space that that redundant files consume. And for some customers this is acceptable. For other customers, this may not be acceptable. The more common methodology is to use deduplication to remove redundant data completely, and free up space for productive use. Dell/Compellent has neither compression nor deduplication, and would have to leave redundant data in place, consuming more space.
Compellent/Dell is a strong incumbent vendor in the mid-range storage market space, which confers upon this vendor advantages in term of some customers preferring to buy an end to end Dell solution. This can in some cases, overcome some of the product limitations in terms of lacking deduplication and compression, and limited protocol support.
Tegile is also a Hybrid Storage Array and similar to Nimble in terms of functionality and design. However, Tegile provides active/active controllers, supports FC, CIFS, and NFS in addition to iSCSI, and provides deduplication capability.
Pure Storage is an All Flash Array, which is performance oriented and is gaining traction in the national Intelligence space. With active/active controllers, dedupe, compression, and a comprehensive software stack, it is nicely functional. But being all flash, it may be constrained in terms of cost-effectiveness for the mid-range market. It is not un-surprising that the Pure All Flash Array is gaining traction in the performance oriented national intelligence space with their deeper pockets.
Tintri focuses on NFS, and runs Virtual Servers and VDI with end to end visualization at a per vm level.
Nutanix is also a niche player, focusing on two key themes: 1) Hyperscale architectures 2) Virtualization. To place this in more concrete terms, their architecture combines both server and storage in one 2 RU (3.5" high) device and is similar in concept to the architectures deployed by Amazon, Google, and Facebook. This is an architecture designed for huge scale, and may be outside the requirements of many mid-range customers. In addition, compute and storage are added in near equal amounts. While consistent with how Hyperscale architectures are deployed, some customers may prefer to optimize for capacity (example: an analyst firm with a lot of data, of which only a small proportion is accessed at one time) or performance (example: a small amount of data that needs to be processed very quickly).
Nimble has validated the Hybrid Storage concept. Financial results in terms of revenue, margin, and unit sales are all in positive territory.
Two key questions are as follows:
1. How far can the stock go?
2. What are the gating factors?
With respect to how far the stock can go, I recommend caution since the stock has already run 130% over its IPO price. The caution is more about avoiding "Euphoria" than having any major negative views on the company or product. There are a few lessons to be learned from the Violin IPO, and one of them is not to get to caught up in the " Violin flash is fast and therefore EMC will be driven out of business" thought process. Clearly, Nimble products are accepted in the market place, but there are gating factors to revenue, margins, and the Nimble stock price.
With respect to the gating factors on the Nimble stock price, here are my views:
1. Addressable market is restricted to the mid-range markets
2. There are several other Hybrid and All Flash Vendors in the mid-range space, and they are credible competitors.
These vendors include general purpose vendors such as Compellent/Dell, Tegile, Pure Storage, and niche vendors such Tintri and Nutanix.
3. While Nimble has done a good job for a small company in bringing a reasonable feature set to market, certain limitations such as active/standby controllers, iSCSI only protocol support, and lack of deduplication provide openings for other vendors to be credible competitors.
4. Caveat: There are many other technical details which are beyond the scope of this financial website.
Appendix A - Requirements for Tier 1 Enterprise
The key capabilities needed to succeed in the Tier 1 Enterprise space include:
RAS (Reliability, Availability, Serviceability)
Data Protection (Snapshots, Copy on Write Snapshot, Clones, Consistency Groups)
Data Reduction ( Deduplication, Compression, Thin Provisioning)
Tiered Storage Management (More suited to Tier 1 Enterprise OLTP)
Disaster Recovery - Asynchronous Remote Replication
Delta Resynch (Re-synch multiple disaster recovery relocations after recovery from failure)
Hot Sites - Synchronous Replication
Security (Role based security, audit logging, internal in-flight, Data-At-Rest), and sometime Common Criteria testing
Partitioning / multi-tenancy
Application Prioritization/QOS (Quality of Service)
Appendix B - Common Use Cases
Use Cases Where Flash Memory Array Start Ups are currently successful
Virtual Desktop Infrastructure
In this use case, desktop images (operating systems and configuration) along with data, are placed in flash memory.
Desktops can be updated and managed at a centralized location, and the flash memory handles a high volume of
reads and writes from the desktops more efficiently than hard disk drives. VDI usually serve workgroups, departments, and divisions.
A flash memory array can serve as high performance storage for many virtual machines residing in a large number of physical servers. The problem this solves is that a large number of virtual machines generate such a large number of random reads and random writes that they can overwhelm hard disk drives. Flash memory, with its ability to handle randomness, is a good fit here.
Analyze data quickly and create reports faster than on disk drives. Reports that once took hours to run can be processed far more quickly. In some (but not all) analytics applications, the only characteristic needed is performance.
High availability, data protection, and a host of other storage management software modules are not needed to be successful. The reason for this is that the initial data capture occurs on a Primary Storage System which is extremely capable, and a copy is given to the analytics application for processing. If the analytics systems corrupts or loses the data, that is acceptable. Because there is another master copy stored safely on the really expensive Primary Storage system, and a copy is reloaded into the analytics system for reprocessing. With that said, there are analytics applications which do require all the storage management software bells and whistles, and it is these class of analytics applications that will pose a barrier to flash memory array vendors that do not have complete or effective storage management software.
SMB (Small Medium Business) Database
Database for small medium business often used for transaction processing. SMB typically do not operate
24/7, and for that reason product technical requirements are less rigorous. Products that are designed to run 24 X 7 with high availability and extensive storage management software stacks are typically far more expensive.
SMB (Small Medium Business) file caching
Hot data is kept in flash memory (the cache) and mathematical algorithms keep track of frequently used data and place frequently accessed data in cache and less frequently used data on hard disk drives. In practice, this is successfully used in smaller installations and the algorithms are less effective on very large data sets (mostly true, there are a few very large NAS vendors with specialized high performance features). File caching is should not be compared to Tiered Storage Management software, despite the marketing claims of some vendors.
Broader List of Use Cases
Online transaction processing
There are two levels here.
Primary Storage used for Tier 1 Online transaction processing (OLTP). This is what large Fortune 500 companies use to process revenue on a corporate wide basis. Examples include large retail banks and consumer retailers. Since money is being processed on a 24X7 global basis for an entire company (as opposed to a department or a division), systems must always be operational and protect he data from loss. RAS (Reliability, Availability, and Serviceability) is top priority and product requirements are very rigorous and expensive to implement.
SMB (Small Medium) database, sometimes referred to as Tier 1.5 OLTP are characterized as smaller operations not running 24X7 and supporting activities at the workgroup, departmental, divisional or small medium business level. Since there is less at stake, technical requirements of the product are less rigorous, and the product is less expensive. In other words, it is easier to enter this market space than Primary storage.
Sharing files such as word, excel, power point, photo, video. This is also referred to as NAS, or Network Attached Storage.
Hot data is kept in flash memory (the cache) and mathematical algorithms keep track of frequently used data and place frequently accessed data in cache and less frequently used data on hard disk drives. In practice, this is successfully used in smaller installations and the algorithms are less effective on very large data sets (mostly true, there are a few very large NAS vendors with specialized high performance features), File caching is should not be compared to Tiered Storage Management software, despite the marketing claims of some vendors.
Analyze data quickly and create reports faster processing data from hard disk drives. Reports that once took hours to run can be processed far more quickly. In some (but not all) analytics applications, the only characteristic you need is performance.
You do not need high availability, data protection, and a host of other storage management software to be successful. The reason for this is that the initial data capture occurs on a Primary Storage System which is extremely capable, and a copy is given to the analytics application for processing. If the analytics systems, fails, corrupts or loses the data, that is acceptable. Because there is another master copy stored safely on the really expensive Primary Storage system, and a copy is reloaded into the analytics system for reprocessing. With that said, there are analytics applications which do require all the storage management software bells and whistles, and it is these class of analytics applications that will pose a barrier to flash memory array vendors that do not have complete or effective storage management software.
Tiered Storage Management (TSM)
TSM migrates data to the appropriate media based on performance, cost, and persistence. For example, move data from conventional DRAM to high performance flash memory, to hard disk, to tape. Data movement is based on policies. An example of a policy is "move all data in Volume 1 from Flash Memory to hard disk drives after 30 days". This is just one very simple example. Tiered Storage Management software is not the same as caching software, and the two software packages fulfill different missions are should not positioned as one being better or worse than the other. It really depends on the customers' business, operational, and technical requirements.
Server Side Virtualization
A flash memory array can serve as high performance storage shared across many virtual machines residing in a large number of physical servers. The problem this solves is that a large number of virtual machines generate such a large number of random reads and random writes that they can overwhelm hard disk drives. Flash memory, with its ability to handle randomness very efficiently is a good fit here.
This is an Amazon, Google, Facebook type of architecture. They use white box Intel-architecture servers and run inexpensive open source software and configuration tools which allow them to manage thousands of boxes and software in an automated manner. The Flash Memory Arrays start-up products are typically not used here; these customers usually use PCIe SSD cards, such as those provided by Fusion IO, LSI, and Intel, and others. Each white box contains both compute and storage.
Disclosure: I am short VMEM.
Business relationship disclosure: I am employed by one of the companies mentioned in the article