RAID is a data storage virtualization technology that combines multiple physical disk drive components into one or more logical units for the purposes of data redundancy, performance improvement, or both. RAID stands for stands for Redundant Array of Inexpensive Disks also known as Disk Mirroring technology.
RAID/Mirrored Architecture in Criticalcase
We’ve chosen Mirrored architecture for our storage because it provides a symmetrical and an exact copy of the data in two different and separate disk piles. The biggest advantages of this type of architecture consist in providing at the same time a great resilience and double the performance.
Mirrored Architecture: Resilience
All the disks in the piles (that are composed of the disks themselves and controllers) are organized in RAID systems, which makes it very difficult to compromise any pile, so even in a case of a breakdown of any pile, the service will remain up&running with no data loss because it’s perfectly and entirely copied in the second pile.
iScsi channels are managed by independent and redundant switches in our datacenters that assure service continuity even in case of a failure of one of them. Criticalcase provide automatic snapshots of all the LUNs present in the Datacore storage with 7 days retention*
*Snapshots do not replace a backup plan, their integrity depends on the integrity of the LUN they are originated on
Mirrored Architecture: Performance
The two piles that share the same copies of the data provide the ser
Both piles are working in active mode, it means that the two piles that share the same copy of data are both assuring service in the same moment. Given the same amount of storage the supply of IO is double. Additionally it is important to specify the controllers are equipped with a great quantity of RAM (minimum 256 GB) that is used as read and write cache allowing the highest possible performance in case of a cache hit. iScsi and mirror channels can guarantee at leas 100Gbit of bandwidth for each pile.
Mirrored Architecture: Scheme
Tiering allows us to store data on different disks and it is an intelligent way to lower the costs without having to give up on the performance. Datacore allows to have an automatic tiering system and it can dynamically move the data in blocks based on the access frequency. Data, that is required more often will be stored automatically on the SSD disks, while rarely requested data will be stored on less performant disks. The algorithm is adaptive.
Storage types in Criticalcase Datacenters
Criticalcase infrastructure is based on 7 European datacenters, Tier III and IV, distributed in Italy and in Europe, all completely redundant. In Data Centers each rack has a dual power supply, a backup system realized by UPS and diesel generators to ensure the continuity of the service at 99.99%.
Criticalcase offers 4 types of Enterprise Storage, all based on Mirrored (RAID) architecture.
The storages are differentiated based on their performance:
- Iron: capacity storage, replaced by SAS disks – advised when the speed is not essential
- Silver: balanced storage without SSD technology. We use 10 RPM capacitive SAS with automatic tiering
- Gold: balanced storage with SSD technology. We use SSD datacenter level and capacitive disks with automatic tiering (50% SSD and 50% capacitive disks)
- Platinum: Full SSD storage with SSD datacenter level
Subscribe to receive precious Tips&Tricks on Cloud Architectures