Technology - Storage
Ultraspeed's Diskless platform is constructed around the concept of highly redundant and scalable, enterprise storage. Ultraspeed is now on its second generation of storage area networks (SANs) and its storage partner is 3PAR, Ultraspeed is also one of the few members in the UK of 3PAR's Cloud Agile programme.
Redundancy. Each SAN Ultraspeed deploys is fully redundant. This starts with the heads (the main processing units that control where disk IO goes among the disks); there are two heads operating in active active mode, which means that data and workload is split between each head. Redundancy is also built in at a cage layer. Data is not just striped horizontally along the shelves it is also striped vertically down across each shelf. This allows the system to suffer a complete shelf or multiple disk failures in one shelf and continue serving data without interruption. In the event of a component failure, the part is either replaced from local spares or under four hours by our partner.
Performance is key in any storage system, Ultraspeed use enterprise storage with fibre channel disks to ensure that there is always enough capacity to allow functional IO to all customers. Ultraspeed monitor IO at an individual disk and customer layer, Ultraspeed also take action in real time to migrate data to other areas of the storage array in the event of more performance being required. This (like all operations and maintenance to Ultraspeed's Storage) is done online with no customer impact.
Scalability is essential. Ultraspeed's storage is designed to scale easily and transparently. The company adds new disks, shelves and heads, according to maintenance schedules, but while storage is fully online and serving customers. Any required software updates to storage are also done in the same way.
Remote replication. Ultraspeed utilises asynchronous replication between SANs, currently between London and Amsterdam.
