Hybrid Storage – Combing speed and capacity
We provide hybrid storage arrays that combine both flash storage and hard disks to provide high-performance storage at an affordable price to deliver the capacity and IOPS a business needs to provide the capacity and performance the software applications require.
Hybrid storage types
Most hybrid storage systems today consist of hard disks for capacity and SSD drives to deliver performance. Seagate manufacture a Hybrid SSHD drive that combines both of these elements and are primarily used in PC’s and games consoles. A hybrid storage system can be a JBOD combining both SSD and hard disks or a RAID system combining a RAID controller. The main issue when using flash storage as a cache or primary storage is that the RAID controllers are not designed to handle 10,000’s IOPS as they were originally developed to handle hard disks and many have firmware adapted to handle SSD. Many of the hybrid storage systems today are designed to scale for capacity by adding additional drive trays rather than increasing performance.
All Flash vs Hybrid Flash
Comparing these is like a Ferrari vs Mini both transport people and that’s where it ends. All-flash is the Ferrari and Mini the hybrid flash, an all-flash array is designed from the controller card, system bus, processors and actual flash storage to deliver maximum IOPS whereas a hybrid flash array delivers far higher IOPS than a comparable hard disk array. Hybrid storage is primarily designed for mixed workloads and flash array for dedicated tasks such as running an Oracle or SQL database.
With disk capacities now at 12TB and rising, SSD storage is at 2TB so hybrid has a capacity ratio of 6x. But many hybrid and all-flash arrays are using data deduplication to achieve far higher storage capacities in order to be more competitive in the storage arena. Deduplication is achieved by examining all of the blocks within a folder, disk volume or raid set and looks for block repetition, then removes duplicate blocks and then compresses the remaining blocks to create a hash # that references the location and quantity of the blocks for recall later. There are many different types of data deduplication source based, target based, inline, post, file or block, fixed or variable length blocks but we are not discussing them here. Data deduplication requires powerful CPU’s and memory in order to rewrite the blocks and hash files. With increased processing power deduplication is becoming more prevalent in many mainstream storage systems and operating systems as costs come down.
How can we help?
We work with the world’s leading providers of hybrid storage to deliver solutions to enable your organisation to grow, deliver data faster and reduce unnecessary expenditure.
If you would like to know more call us on 0207 193 5760 or complete our form below.