THE DDP VISION
The magic words are hybrid DDPs with SSD caching and adding DDPs on the fly with automated data redistribution. This can be done without administrative changes in the web interface.
And there’ll be more magic to come with the Here, There and Elsewhere technology. With this technology even any DDP can be added on the fly - again with no administrative changes.
SSD Caching, Load Balancing, Mirroring and Linear Scaling
The data is separate from the file system and is stored in data container called Data Locations. This allows file based SSD Caching and Load Balancing. Originally Data Locations selection is within one DDP. With the latest software Data Locations can also be other DDPs or parts of other DDPs, thus building a DDP Cluster. A DDP Cluster scales lineary in bandwidth and capacity, so called scale out. Also two DDPs can be mirrored.
Why is SSD caching with the DDP more practical than with other systems?
DDPs are modular and can hold HD and SSD packs of various capacities. Caching is between SSDs and HDs as Data Locations and is file based. The SSDs as Data Location can be selected as cache. Data Location selection is foldervolume based. When the SSDs selected for caching files are transferred in the background transparant for the operators. The SSD Data Location can also be used as primary storage. Older DDPs can be retrofitted with an SSD pack as well.
Why does Load Balancing result in higher performance?
Load Balancing involves multiple Data Locations and is file based. It is designed to compensate for the seek time of hard disks. Data Locations are accessed independently. The more Data Locations there are the more independent accesses there are and the higher the total throughput will be.
What is special about the DDP?
Previously we discussed the major features and functionality of the DDP. Under the hood there is a virtual file system called ddpvolume, the so called metadata.
The data is separate from the file system and is stored in data container called Data Locations. Data Locations can be within a DDP but can also be other DDPs or within other DDPs.
Why does a DDP Cluster scale linear both in capacity and bandwidth?
DDDPs can be added as Data Location with each desktop having parallel access to all DDPs. Two identical DDPs double throughput and ten increases throughput tenfold, although DDPs do not need to be identical however. As an example one microDDP10GbE has a bandwidth of 2,2 GB/s. And ten have a bandwidth of 22GB/s.
The DDP Cluster
The variety of DDPs shown is part of one file system: ddpvolume, one namespace. Not only can the DDPs differ, they can also be of different build dates. Data within a Folder Volume can come from one Data Location or be balanced between a group of Data Locations, e.g. a group of DDPs. In the background without changes in directory/folder/file tree structure within the DDP cluster, data can be transferred between Data Locations. Also data can be redistributed between Data Locations or consolidated on one.
DDPs can be removed from the cluster, used standalone and added on the fly again. Without copying the data is immediately available within the cluster.