Avere – Can Automated Tiered Storage Reduce Our $120 Billion On Storage?

We’ve come across a number of highly innovative smaller companies over the last couple of years – companies like Aaxana and Racemi who have new approaches to the normal way users implement systems resulting in significantly lower costs. We travelled up to the Royal Exchange in London recently to talk to CEO Ron Bianchini and VP Marketing Rebecca Thompson of Avere – a small storage systems company with a compelling story. We thought it would be interesting for our readers to learn about them. We also want to share some of our sizing of the storage market and look at how Avere’s approach might reduce it.

Who Is Avere?

Ron and his team founded Avere in 2008. He has a lot of relevant technical and commercial experience, having been at Spinnaker Networks, which developed the Storage Grid acquired by NetApp, where he then worked as head of the Pittsburgh Technology Center. In a similar way earlier in his career he founded Scalable Networks which designed and implemented a large-scale Gigabit Ethernet switch, before being acquired by FORE Systems, where he became head of ATM products.
Rebecca’s career started in market analysis at Dataquest, before moving on to run analyst relations and marketing programmes at Cisco, FORE, Freemarkets and Vivisimo. We had some interesting discussion about our analyst colleagues from those days.
Avere is based in Pittsburgh. It is a privately held company, having gone through two venture capital funding rounds. It currently has around 60 employees and is active in the US, UK and France, selling under its own name with no OEM business.

How Does Avere’s FXT Cluster Work?

Avere’s skills are in advanced automated tiered storage, based on its A3 technical architecture. Its FXT 2700 appliance is based on a Supermicro design and built by and Arrow. It’s a NAS appliance to be installed between servers and a company’s filers (see Figure 1). By automatically equivocating between tiers 0 to 2 it reduces the amount of data going out through the filers to disk systems by an average factor of 55:1. Of course users have to pay for something new (i.e. the appliance), but make savings through the reduction in capacity and the use of tier 3 SATA as opposed to SAS and Fibre Channel disks. The Avere box swaps between tiers in real time at the block – rather than volume – level.
When compared with the automated tiering approaches of HDS and IBM Ron claims his solution moves much less data around and needs far less transactions. FXT uses NFS V3, making it relatively vendor agnostic in terms of the storage attached at the backend through the filers.

Testing its solution against NetApp and EMC with the sfs benchmark for 100 ops/second systems identified its cluster as having significantly lower costs both for ops/second ($3.4 v $11.3 and $76.3 comparatively) and the overall system ($445k v $1.351m and $8.435m comparatively). Despite its appliance packaging FXT can be expanded via scale-out clustering and Ron claims this is better than with NetApp, where upgrade paths are limited by model ranges. Ron’s networking and clustering experience allows Avere products to be added together to create storage pools effectively.

We Spend $120 Billion On Storage

The amount we spend on storage can be broken down into a number of different categories. In particular:

  • Revenues from disk drive manufacturers in the year to the end of June 2011 were $33 billion
  • The storage systems market was worth $53 billion over the same period
  • The world server market was worth $63 billion and PC market, $183 billion, with storage accounting for a large proportion of both

All things considered and taking out the revenues spent by one supplier with another for components and OEM products, we estimate the total users spending on storage at around $120 billion worldwide.
Reducing storage is about increasing the utilisation of devices and reducing the amount of data that gets recorded by throwing unnecessary data away. Symantec, EMC and others suggest de-duplication as one means of doing this of course, while the grand shift towards centralisation also cuts the need for storage through better utilisation. Avere’s approach is an extreme instance of using automated tiering techniques to reduce the reliance on under-utilised expensive drives and systems.
Driving storage demand ever upwards is the digitisation of analogue processes and the capturing of new monitored and metered data. IBM’s Smarter Planet initiatives, for instance, see Cities and other large organisations handle massive amounts of new data: while the increasing use of feeds from social media interaction is also expanding storage demand. De-duplication, automated tiering, consolidation and centralisation will undoubtedly reduce costs, but do little to hold back the torrent of data entering and being handled by data centres.

The Increasing Demand For Disk Systems

Around 200 million PCs and 6 million servers were sold worldwide in the year to the end of June 2011. Our investigation into disk drives suggests that 970 million were shipped over the same period (for a quarterly view see Figure 3). Despite the variety of product types available (more later), the average price of disc drives reduces over time. Our research suggests that the average cost of each disk drive was $57 over the last year.

Different Disk Types Offer A Vast Variation In Cost And Capacity

The exponential growth in data demand is somewhat limited by the industry’s ability to deliver drives and systems at affordable costs, which in turn is controlled by the development of new devices with smaller form factors, greater capacities and massive manufacturing scale economies.
Avere’s research illustrates the extent to which the cost per GB increases and capacity decreases as you move from one level of storage to another (Figure 4). It exploits the differences by keeping the more expensive higher-tier drives highly utilised in its appliance, offloading a much smaller amount of the lowest tier through the filers to SATA devices.

Some Conclusions – Avere Should Build A Good Business

Avere’s solution is innovative and clear in its value proposition. Its success isn’t limited to benchmarks, as it also has good customer references such as ION in the Seismic Oil and Gas area and Sony Imageworks.
It proposes to lower user spending on storage significantly through applying its technical knowledge. It is early in addressing the disk cost issues which will become increasingly urgent as the mass of new data requires huge amounts of disk.
Netazza, Compellent, 3Par and other NAS specialists have all now become part of big systems companies, allowing them to modernise their approaches and address the developing market for Big Data, however defined. Whether Ron follows his own (and their) history to be similarly acquired or chooses to build the Avere brand, we expect to see significant adoption of these products in future.

Please contact us if you have ideas about this or other innovative approaches to improving performance and/reducing costs.

2 Responses to “Avere – Can Automated Tiered Storage Reduce Our $120 Billion On Storage?”

Read below or add a comment...

Trackbacks

  1. […] late-night or other discounted schema. Like other start-up storage companies (Tegile, Xio, Avere, Violin Memory, etc.) SolidFire has the challenge of persuading customers to trust their priceless […]

  2. […] took a look at Avere in October 2011, to weigh up the advantages of using its edge filer and in November 2013 to consider its invention […]