Storage Area Network
Introducing SAN technology in a hosted environment brings many advantages. In short, a SAN will provide storage capacity to systems in such a way that, to their operating systems, the storage appears to be a local disk. These ‘disks’ are however nearly unlimited in size and easy to manage, backup and copy. Among the benefits are improved storage utilisation, high availability and data protection.
Fibre Channel versus iSCSI SAN
Today there are two protocols available for building block-based SANs; FC (fibre channel) and iSCSI. Both protocols use the SCSI commands generated by the file systems of the servers. These SCSI commands are then converted by the iSCSI or FC protocol so they can travel via a network to and from a centralised data storage system (typically one or more large disk arrays) where the commands are executed. In the case of FC, the network equipment (cabling, switches, and interfaces) is specific to the protocol. In the case of iSCSI the network equipment type is anything that will handle IP packets, 1GB Ethernet being most popular.
iSCSI-based SAN has become a serious option for environments that must have a form of central storage, as iSCSI uses standard interface cards, tools and the standard knowledge on IP networks within their IT departments. Yet do note that – due to the fact that data is being read and written via an IP network instead of to/from internal disk drives – users commonly are concerned that latency will degrade the net performance.
Terremark has been using both SAN types and has thus built up thorough knowledge. In fact we were among the first to test and deploy iSCSI SAN technology, driven by our customers’ growing storage needs. Among the measures we take to optimise performance of iSCSI storage are the use of jumbo IP frames and the use of dedicated storage networks (not shared with any other traffic type).
Network Attached Storage
Similar to SAN, Terremark deploys NAS technology to deal with large amounts of data. A characteristic difference between SAN and NAS is the way they provide their storage to the attached systems. SAN provides ‘virtual file systems’; the attached servers simply perceive a local disk set (which actually is part of the central SAN storage capacity).
NAS however presents its storage capacity on a network-level, meaning that the systems within the network can access one or more network shares that reside on the NAS. In this respect, the NAS device simply acts as a central file server.
This being so, there are limitations to what the servers can do with that storage capacity. For example: no two systems can simultaneously write to the same file. Based on its extensive experience with NAS technology, Terremark can deploy supporting services (in this example we would use NAS heads to solve this particular situation).
Media Content Storage
Media Content Storage (MCS) is based on ‘Content Address Storage’, or CAS, which is a data storage method specifically aimed at storing static files (i.e. objects that will not change). These include photographs, videos, digital archives, online forum archives, etcetera. By assigning a permanent place within the storage farm to any object saved, access to it is simplified (allowing for easy integration with your applications) and is very fast.
The technology furthermore prevents any object from being saved twice, thereby reducing the costs involved with excess disk usage by files that have been saved more than once. And since Media Content Storage is comprised of low-cost server farms instead of costly technology, the costs per stored gigabyte remain attractively low.
Terremark in 2008 already deployed this technology, for instance to serve a publisher that runs an online video clip service (similar to the YouTUBE concept). The fully managed MCS service reduces costs for that publisher, is fully transparent for the publisher and the end user, and substantially simplifies the way that the data is managed.