I’m very fond of Object Storage but, when we come back to real life, its adoption in ordinary enterprise environments is slowed down by many factors. On the other hand there is a strong demand for more efficient, scalable and robust storage solutions to cope with new enterprise data challenges and data growth is only the tip of the iceberg.
Amazon S3 as the point of reference
Amazon S3 Object storage service is unquestionably one of the most successful cloud services ever and the recently introduced Glacier (for long term archiving) is set to become another de facto standard. The S3 service was primarily designed as a storage for web based applications hosted on the public cloud, but it’s clear that it is quickly becoming a viable backend for many other solutions.
Enterprises are well aware of this. They are looking at this phenomenon with interest and, sometimes, the idea to replicate the same successful platform internally.
This is also partially related to the visible change that we are seeing in the way enterprises are looking at the transformation of their IT and new delivery models: private clouds and mobile devices are only the easiest to mention.
2014
This is why 2014 will see a substantial growth of these kinds of solutions.
In any case, next march I’ll be attending Next Generation Object Storage Summit in LA. An event that will host most of the Object Storage vendors, analysts, press, industry pundits and end users. It is going to be a great opportunity to join the conversation and stay updates on the evolution of this market.
In the mean time, the purpose of this post is to give the reader an idea of the available object storage platforms out there and I’ll try to spend a few words on each one.
As I have recently written here many times, to have a successful Object Storage strategy you need a complete ecosystem:
– A horizontal platform providing capacity, resiliency, scalability, data protection, replication, standard APIs, policies and management features.
– A number of gateways to translate traditional common services into this new way of storing data.
In this post I’m going to cover only the first part, leaving the gateways for a future post.
Just a short disclaimer
A comparison between Object Storage systems is even harder than comparing traditional arrays so don’t take this as a scorecard or a competitive sheet. Not only because the features and technology are differently implemented but, in most of the cases, products are originally designed with different objectives in mind. My intention here is to give a list of references and a glimpse of the spectrum for the possible solutions.
Amplidata Amplistor
Amplidata was founded in 2008 and the product is called Amplistor. This is a hardware solution based on standard x86 boxes (intel is an investor here).
The front-end nodes are dedicated to all the IO and CPU intensive tasks while the high-density backend nodes provide space. Due to the data protection algorithm used in this solution (erasure code) and its implementation, the minimal possibile configuration is larger than needed for most enterprises.
This solution is very focused on large files and big environments (like video archives for example) limiting its adoption in the ordinary enterprise. The product is also not very well supported by gateway and software vendors, leaving a short list of choices to implement a complete architecture.
Quantum Lattus Object Storage is based on this product, someway confirming the media&boradcasting vocation.
Caringo CAStor
Caringo was born in 2005, founded by a team of people who have a long career in developing technology and solutions in this market segment. CAStor is probably one of the most widespread platforms out there, counting more than 500 installations.
Technically speaking this product has some interesting characteristics that allow customers to scale from a very small cluster to multi-petabyte installations. The way CAStor manages objects and data protection at the backend allow to cope with small and large files at the same time. Caringo also offers some companion products to enable access to CAStor not only via APIs but also through a scalable NAS front-end.
It’s sold (directly and indirectly) as a software only solution (easy to deploy in virtual infrastructures too). In the years Caringo has built an interesting number of relationships with ISVs and solutions providers giving the end user an interesting freedom of choice for deployments.
Cleversafe dsNet
Cleversafe is a company based in Chicago, founded in 2004 that is developing a product called dsNet. This is one of the kids on the block of modern Object Storage, very well founded and with one of the most interesting solutions at the web scale.
Cleversafe primarily targets huge geo distributed installations in the range of multiPetaBytes to Exabytes, leveraging one of the best erasure code implementations of the market as well as well designed and reliable security mechanism. The architecture is, in certain aspects, similar to Amplidata: front end nodes that manage most of IO and high capacity storage nodes at the backend. Due to the nature of the solution, and related issues in managing small files, Cleversafe has recently enhanced its product lineup with high performance (SSD) nodes for caching and speeding up certain type of operations. The management system is very well designed and allows to manage giganormous installations quite easily.
The product is normally sold on preconfigured appliances but Cleversafe can offer the software alone in some circumstances (I recently found this at HP Discover). Cleversafe offers only API access but has partnerships with other vendors to provide certified NFS/SMB access gateways as well as sync&share solutions. They are clearly positioned to serve high-end needs.
Cloudian
I don’t know Cloudian well, and the only things that I can mention come from their website. I hope to meet them soon and fill the gap.
Cloudian provides a full Amazon S3 compatible Object Storage based on a common scale-out architecture very similar to what you can find from other vendors.
The only feature that they publicize on their website is the S3 compatibility, a functionality that is available from most of the object storage systems today. So, at this point I can’t really find a differentiation in their offer, even though on their website there is a long list of partners and compatible solutions to build a complete ecosystem.
See you next week
Next week I’m going to cover other major players of the market (Data Direct Networks WOS, EMC Atmos, Hitachi Data Systems HCP, Inktank Ceph, NetApp StoreGRID, Openstack Swift and SwiftStack, Scality RING). Stay tuned!
Ps: Sorry if you can’t find your name in this list but these are the names that are visible in the market at the moment and with the broadest attention from end users. If you think that you should be mentioned here let me know, I’ll be very glad to set up a chat with you.
Excellent article. I am desperate to see the rest of it. Especially where EMC and DDN fit into the mix. Do you have any insight into how well the object storage vendors integrate with the standard backup software platforms? Tivoli, Symentec, Commvault?
Brian,
Thank you for the comment!
Object storage and Backup has some interesting points of contact.
For example Commvault can use S3-like repositories as a target and it is also has (or it will be released soon) a S3 interface to access backup data.
Others have similar approach and the number of solutions is quickly growing.
Enrico
I can speak to Hitachi Data Systems’ Hitachi Content Platform (HCP). With TSM, HCP is a Sequential Access Storage Pool like tape devices, optical devices, files on disk or storage on another server. With CommVault, HCP can be connected via NFS, SMB and, ideally, via HTTP as a Cloud Library. In the case of Symantec, HCP is tightly integrated with their Enterprise Vault solutions, HCP also provides an S3 interface that can work with NetBackup or BackupExec. I hope this helps.