Amazon Re:Invent Briefing – Igneous – S3 in your Data Center

Despite the ever increasing acceptance of the cloud, an undeniable reality is most organizations will have data they will never put into the cloud. At the same time these organizations “get” the value of an elastic storage infrastructure that they don’t have to manage. But for a provider to deliver storage as a service to a single location creates economic, performance and support challenges. Providers need to overcome those challenges to offer a competitive alternative to the cloud or traditional on-premises storage. Igneous has a unique S3 compatible storage solution that it feels overcomes all of these challenges, allowing it to deliver on-premises cloud storage that customers buy in a similar fashion to public cloud storage.

Meeting the On-Premises Storage as a Service Challenge

All of the challenges an on-premises storage as a service solution faces center around one factor: lack of scale. One data center does not typically allow the traditional cloud storage design to reach enough scale for it to meet the performance and cost expectations that IT professionals have of cloud storage.

The problem is lack of granularity in scale and protection. Most private cloud (object or S3) storage systems are a cluster of individual storage servers, called nodes, with internal capacity that are aggregated into a single storage pool. When more capacity or performance is needed, administrators add an additional server to the cluster. But that means each expansion needs an entire server, which is likely overkill if you need only a small increase in either performance or capacity. It also means the failure of a one or two drives within that server impacts the entire server thus bringing it offline. The cluster then loses all of the remaining capacity and performance that the server had available.

From an “as a service” perspective the lack of granularity means the provider will need to dispatch personnel too frequently. Not only is this is a cost issue but also an access issue. Remember in this use case, the provider does not have full access to the data center like a public cloud provider does to its service.

img_20161130_150534

The Igneous Solution

Igneous meets these challenges by making the point of scale more granular. As you can see in the pictures, they put the compute ON THE DRIVE. That means each drive has its own compute and ethernet connectivity. This provides plenty of performance and bandwidth for the customer and allows Igneous not to have to deploy a service technician every time a few drives fail within a server. Igneous uses JBOD chassis to store these “nodes” and each drive is independent of the other drives.

img_20161130_150517

Rethinking Hybrid Cloud

The Igneous system is native S3, which means that any storage operations that work on it will also work in the Amazon cloud. Igneous essentially re-positions how an organization should look at cloud storage, instead of cloud storage being a final destination point, organizations should look at it more temporal. For example, if there is a sudden spike in capacity needs IT can store data in the cloud until it gets more on-prem capacity. Or even more interesting: data can be pushed to the cloud to leverage cloud compute and then pulled back to the data center after completion of the task.

StorageSwiss Take

Igneous is solving a problem that we think has plagued scale-out architectures for quite sometime: Lack of granularity in the way they scale. The ability to make each drive in the architecture essentially a storage node should deliver a massive performance improvement and allow Igneous to offer a very cost competitive storage as a service solution.

Eight years ago George Crump, founded Storage Switzerland with one simple goal. To educate IT professionals about all aspects of data center storage. He is the primary contributor to Storage Switzerland and is and a heavily sought after public speaker. With 25 years of experience designing storage solutions for data centers across the US, he has seen the birth of such technologies as RAID, NAS and SAN, Virtualization, Cloud and Enterprise Flash. Prior to founding Storage Switzerland he was CTO at one the nation's largest storage integrators where he was in charge of technology testing, integration and product selection.

Tagged with: , , , , ,
Posted in Blog
3 comments on “Amazon Re:Invent Briefing – Igneous – S3 in your Data Center
  1. Tim Wessels says:

    Well, there is so much to argue about with Mr. Crump’s blog that it is hard to know where to get started, but his first sentence is correct. The rest of the first paragraph reminds me of the ill-fated storage provider Nirvanix. Three years ago Nirvanix was providing on-premises private, hybrid and public cloud storage to customers while trying to run with the big dogs of cloud storage. The problem was the Nirvanix business model was flawed due to the capital investment they had to make on each customer’s behalf. Eventually working capital became an issue. Then there was a year of executive churn. Then a frantic search for additional funding. Then Nirvanix shuttered their storage service on short notice after blowing a $70M hole in the ground. Oracle picked the bones of the Nirvanix carcass a few months later.

    In the second paragraph, Mr. Crump fails to recall that storage server nodes in an object-based storage cluster can be of any size. The object storage software is hardware agnostic, and the storage server nodes can be heterogeneous regarding capacity and processing power.

    It would be interesting to know if Igneous is using a Seagate Kinetic-like HDD, for its non-storage server approach to object storage. Their website makes no mention of it. AFAIK, Kinetic has not reached production scale deployments three years after Mr. James Hughes from Seagate introduced Kinetic. WD/HGST was exploring the use of Ethernet enabled HDDs with Debian running on each drive which is technically not the same thing as Kinetic HDDs although they both use dual Ethernet interfces on each drive.

    Of course all of these Kinetic or Ethernet HDDs plug into a large backplane with up to 90 HDDs in a single 4U enclosure and inside each enclosure is layer 2 Ethernet switching hardware. So, if you lose the chassis power supplies or your Ethernet switch dies you can can take up to 90 “independent” drives off-line at once. BTW, each drive in a JBOD chassis connected to a storage server node(s) are also independent drives.

    So Igneous has some level of S3 API compatability, which is table stakes for every object-based storage sofware vendor. But does this mean you can take the AWS S3 SDK and write an application that will run against the Igneous cluster? Their website is a little vague on how all of this works. And as far as “pushing and pulling” data to and from a public storge cloud on demand, you better have high-speed Internet connections that won’t leave you waiting and this can cost a pretty penny each depending on where you live.

    Igneous looks interesting but their website lacks useful information. I will remain a skeptic until they are more forthcoming about how it all works. I wonder if Mr. Crump has actually seen it all up and running?

  2. George Crump says:

    Tim,

    While I appreciate your participation on our site. Responding to a blog that I wrote at and event, where I was merely reporting on what I saw, claiming that I got things wrongs is a bit over the top. Also responding with a comment that is 3X longer than the post is a bit extreme. I was just writing about what I saw and then giving my first impressions. I suggest that you tempor your comments and try to understand the purpose of the posting. I no where suggest that Igneous replace everything you own. BTW the obviously are not using the Seagate drives as the pictures clearly show.

    George

    • Tim Wessels says:

      Well, I’ll grant that you did label your blog post as a briefing. Sometimes I just keep writing, so I’m sorry if it was overly long. I’ll try to be pithier with my comments. Yes, I did see the add-on board in the photo. I believe Seagate also had similar boards built to modify a standard SATA drive for use as a Kinetic drive. Back when Seagate announced Kinetic, I was very interested in what they were doing. Mr. James Hughes from Seagate gave a terrific presentation on Kinetic at Basho’s RICON West conference in October 2013. You can still find it on YouTube. Unfortunately, Kinetic has yet to live up to its promise.

Comments are closed.

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 21,768 other followers

Blog Stats
  • 1,083,686 views
%d bloggers like this: