Greasing the Skids on Object Storage

Anybody heavily entrenched in the storage aspect of IT infrastructure is excited about object storage these days. And for a good reason. Data for some of the biggest cloud-based services available today is often stored as objects. Using an object storage solution allows for massive scale and durability because there’s no filesystem overhead to deal with. It can be cheaper in comparison to file or block when dealing with significant scale. There are functional benefits as well; for instance, the metadata describing the object also provides rich information about the stored data – something block storage doesn’t provide.
I remember a report from a while back (I can’t seem to find it now) wherein IDC estimated that data storage needs will have a CAGR of 42% through the year 2020. Whether this is accurate or even if it’s half of that, organizations are going to need to find ways to store lots of data. Object storage is a tool in the IT architect’s tool chest to aid in serving up this tall order. However, there seems to be a bit of a disconnect between the exciting stories most object storage vendors are telling today and the rest of the addressable market.

Who is Object Storage Good For?

At Storage Field Day 11, almost every presenter talked about the applicability of their solution to the Media and Entertainment vertical at length. Now, M&E is a 1.5 trillion dollar industry, and some estimates suggest that this market segment alone will be storing 100 exabytes of data within the next year or two. M&E applications and workflows have some unique and aggressive demands concerning both capacity and performance. Make no mistake about the fact that M&E is a vertical that has a great need for technology like object storage.
In a presentation with either “object storage” or “big data” in the name, one is also likely to hear about applicability to life sciences, healthcare, and FinTech. And again, these industries have needs that are a perfect fit for technology like object storage. But what about everyone else? Does object storage have value for the average enterprise or even medium-sized organization? And if so, what options are available to them?

Avere Systems presenting at Storage Field Day 11

Object Storage in the Enterprise

Many customers are exploring the idea of how they might be able to “do more with less” or save on some direct costs with object storage. And for some organizations, there’s a reasonable case to be made for the adoption of a trusted on-premises object storage system like one from Cleversafe (now part of IBM), Amplidata (now part of HGST), or Scality. Sometimes object is also an enabler: for example, DSSD’s underpinnings rely heavily on object-based storage. There’s a significant enough cost savings or the scale requirements mandate a new approach, so CIOs dedicate the engineering time required to get their object storage strategy into production.
But for others who may be interested, there’s a slight problem: the entry point from a capacity standpoint from these reputable object storage vendors is large. This constraint makes sense – after all, object storage is most suited to handling massive amounts of data. The challenge for many would-be adopters is that the entry point for many of these systems is expressed in petabytes. That’s more capacity than a good chunk of the market has in their entire data center, and it’s certainly more than many of them are going to pick up to just get started working with object storage. Without building something homegrown, it seems to be relatively difficult to test the waters of object storage; I get the sense that you have to jump right in the deep end. Not just from a technical perspective, but from an investment perspective.
Of course, this is only a problem if the organization is interested in deploying on-premises object storage. When there’s an insurmountable barrier to entry there, they can just begin consuming object storage on a pay-as-you-go basis from a public cloud provider instead. The most widely known option is Amazon S3, and there are many products in existence to augment your S3 experience by accelerating it and extending Amazon S3 down into your data center. That’s all well and good, but for various reasons, keeping data on-premises is desirable to some organizations. Without a viable on-prem entry point and an aversion to public cloud consumption, it can seem like these folks are out of luck.

There’s Another Way

At SFD11, we heard from Avere, who is well established in the area of accelerating NAS and more recently in enabling hybrid cloud strategies. For organizations looking to disaggregate infrastructure resources and run some on-prem and some in the cloud, Avere certainly has compelling options. However, the last bit of their presentation is what caught my attention.

C2N Architecture
The C2N architecture: FXT Edge filers for performance, CX storage nodes for capacity, and intelligent tiering between local NAS and Object tiers as well as public cloud object tiers.

Recently, Avere brought to market the C2N solution (Cloud-Core NAS). This solution is a combination of NAS and object storage that has two major advantages as I see it:
Primarily, it’s configurable to as small as 120 TB usable capacity. That is well below the entry point for most object storage solutions. Upon system initialization, a customer can choose to use triple replication for the smallest systems (which causes a relatively hard hit on usable capacity) or they can opt to use 8+4 erasure coding for larger deployments (where usable capacity is more like 66% as opposed to 33%). The system does scale to up to 5 petabytes, meaning that if you start in the hundreds of terabytes range, you’ll have plenty of room to grow.
Secondarily, the POSIX-compliant front end can help organizations start getting some of the benefits object storage can provide while still leveraging file access methods they’re used to using. Going straight to object-based access methods is a difficult – if not impossible – road to cross for many customers; C2N provides a helpful bridge.

My Thoughts on C2N

As interesting as the demanding requirements of M&E and the other hot storage-at-scale customers are, there’s still an entire medium & enterprise market out there that will soon have the same sorts of needs the cutting edge industries have been having recently. The most proactive and progressive of them will be looking to start consuming some object storage immediately, and the C2N offering is a great stepping stone for helping those customers get started.