Object storage is changing the data center. Commodity storage offerings provide a well-performing alternative for...
expensive proprietary SAN filers. In this article, you'll learn more about what exactly object storage is, about different object storage products -- such as Ceph object storage -- and whether or not it is something you can use in your environment.
Object storage comes from cloud environments and is based on an architecture where many storage nodes work together to deliver data in parallel, which is fast. We're not talking just a couple of storage nodes -- most object storage products consist of dozens and, in some cases, even hundreds of storage nodes. When a client requests access to the storage, the binary objects, which make up the user data, are quickly delivered to the user because they are delivered in parallel.
Using object storage brings several benefits. Compared to proprietary SAN offerings, object storage is fast, scalable and inexpensive. The biggest savings are in the expense on disks. If you use a filer from a SAN provider, you'll only be able to install top-range serial-attached SCSI disks that cost dollars per gigabyte. In contrast, you can use a commodity disk with object storage. Since data is delivered by many servers in parallel, you don't need to use the fastest disks -- commodity Serial Advanced Technology Attachment disks will do, as the speed is in the network.
There are currently three different products for object storage dominating the market: the legacy Swift, Amazon Simple Storage Service (S3) and the more recent Ceph object storage offering. Swift is mostly used in OpenStack cloud environments and works with applications that address Swift object storage through direct API calls. That means it's fairly limited in use: If you have a generic application or OS, there's no easy way integrate with Swift.
S3 has been around for a long time and works in Amazon cloud environments. Its access methods are limited as well, which means it's not the best candidate for a generic object storage product. S3 is best used to deploy images in an Amazon Web Services cloud environment. Unfortunately, this isn't helpful if you're using VMware vSphere.
Ceph is the most open of all the object storage offerings, not only because it's open source, but also because it offers several different client interfaces:
- API access. This is the most common access model in object storage, but it doesn't work for VMware environments, as you would need to rewrite the vSphere code to access it.
- The Ceph file system. This is a special-purpose file system that can be used on the object storage client. Since this object storage client would be an ESXi server, this option also isn't very usable in VMware environments.
- The RADOS Block Device. This adds a block device to the client OS by loading a kernel module and integrating it on ESXi; this is also difficult to use in a VMware environment.
- The new iSCSI interface. This is a new and promising development in Ceph object storage. In the new iSCSI interface, the Ceph storage cluster includes an iSCSI target, which means the client will access it like any other iSCSI-based SAN offering.
Of these four access methods, the iSCSI interface is the only one that really works in a VMware environment. You may be wondering, doesn't that just replace one SAN product with another? The answer is absolutely not. Even if the client only sees an iSCSI target, you'll be dealing with a flexible, scalable and affordable SAN offering on the back end, which is much cheaper than traditional SAN environments.
The iSCSI target interface for Ceph object storage is relatively new, and you'll notice it may not be available on all Ceph object storage products. It is included in Ceph's SUSE-supported offering, SUSE Enterprise Storage 3, and it is likely that other Ceph vendors, such as Red Hat, will soon follow suit. The iSCSI interface code shows in SUSE first because SUSE is its main developer.
Since Ceph object storage is revolutionizing the world of enterprise storage, it might be a good idea to take the time to explore its possibilities, especially in VMware vSphere environments. Once configured, it will behave just like any other iSCSI data store.
Major improvements come to Red Hat Ceph
The top object storage startups of 2016
Comparing cloud versus in-house object storage