If IT had a mascot, it would have to be the pack rat. Let’s face it; we don’t delete data anymore, and this situation is unlikely to change. Stumbling under the weight of massive data growth (and the expense of traditional storage options to deal with it), lots of organizations come to us looking for better options. They want to cure themselves of the management headaches that come with hundreds of terabytes or petabytes of data, and they want to stop paying big money when they don’t need to for their unstructured data.
When we hear these frustrations from clients, we think of two words: Object. Storage. Unlike block and file protocols, object storage offers more or less infinite capacity. There are few companies on the planet that would come close to pushing the limits of scalability with object storage.
There have been drawbacks to this option in the past few years – you need a different set of tools to access your data when you store it this way, and those tools weren’t widely available until recently; there haven’t been good ways to tune performance. IBM’s Cloud Object Storage, formerly known as Cleversafe, addresses both of those previous concerns, while also offering two compelling benefits for putting unstructured, sedentary and expanding data onto an object storage platform.
Why IBM Cloud Object Storage?
- Erasure coding: Regular storage is built for obscene performance and reliability (and this is one reason why it’s so challenging and expensive to scale). Object storage has traditionally addressed reliability through replicas. In order to achieve scale, object storage doesn’t use metadata. You would write the information and get a unique code or identifier for your object, and then create two or three replicas of it to meet reliability needs.
Not only was this terribly inefficient (1 gigabyte of data ate up 3 gigabytes of storage once you copied it in triplicate), but it also posed another problem. If you lost that code or unique identifier somehow, you also lost your data. IBM solves the first issue by dispersing with erasure coding. Erasure coding is similar to RAID parity, but slower and more flexible. You can define your erasure code algorithms to reflect the level of redundancy you need for your data and balance that number against your efficiency goals. A small tweak of even 12 to 15 percent toward efficiency can have an outsized effect of hundreds of terabytes, which makes a real dent in a pack rat’s storage burden.
- Security: IBM Cloud Object Storage supports out-of-the-box encryption. This is a significant issue for organizations in a growing number of regulated industries. IBM supports encryption for data on its object storage platform, and that fact alone makes object storage much more appealing to a wide range of clients. In addition to encryption, Active Directory or LDAP authentication is supported so it can easily integrate into an enterprise environment. And finally, object-level access lists (ACLs) are available with S3 API compatibility for granular access control.
There are now many tools available that make object storage easily accessible, and there are more tuning options for performance needs. With these shifts, IBM has done a lot to make cloud object storage a cost-effective and attractive option for those facing petabyte or exabyte-level scalability needs.
Learn more about how to solve enterprise storage challenges.
Director of Cloud Services
Key Information Systems, Inc.