Reliability, autonomy and replicas
Scality says every node constantly monitors a number of its peers, presumably the ones in its segment, and automatically rebalances replication and load among them to make the system self-healing if a node crashes. The same mechanism can be used to cope with nodes jointing the ring as it grows.
When an object is first loaded, the Accessor node involved may assign a storage class to it which can define the number of disk failures it can survive: one, two or more. That implies that for it to survive one disk failure, there need to be two copies and so on. The Ring manages the number of replicas requested for each object.
The Ring also produces reliable performance, roughly equivalent to a storage area network, although it would be interesting to see what happens with a 10,000-node ring (four hops) and a 100,000-node one (five hops).
Lecat said: "The number of nodes is theoretically infinite since complexity increases less than the number of nodes. We are very confident about 1,000 nodes, but we have not been able to access enough hardware yet to actually test 10,000."
The Ring dumps low activity data off to a second tier of storage and the data sent there is compressed and deduplicated.
Customers and funding
Cloud service providers use Scality's Ring to offer storage-as-a-service applications, such as email. These customers are European, Scality not being a Silicon Valley-based startup, and include Belgian broadband cable service provider Telenet; Host Europe, which offers cloud hosting; German cloud computing providers ScaleUp and Dunkel; and German web hoster intergenia.
Revenues from these companies explain the relatively low funding level of the June 2010 A-round, in which Scality received $5m funding in June from three French venture capital concerns. This followed on from seed financing of $1.3m in February, when the firm started by morphing out of Bizanga. After that, the firm sold a Mail Transfer Agent product to Cloudmark and looked for a new business to enter. Much of this seed finance came from Scality's employees.
Scality has partnered with CommVault and CommVault Simpana can now be provided in the country of operation by local cloud storage providers using the Scality RING platform, or on-site for customers wishing to deploy their own private clouds with the same technology from Scality. That gets over an issue of legal and social needs for cloud-held data to be stored in the country of origin.
So what do we have here? Scality is an unusual creation: a France-based startup with up-to-the-minute object storage technology that is actually in use by service providers, billing customers real money for using Scality's Ring. They could have decided to use Caringo, Atmos or Centera from EMC, CleverSafe, or HCP from HDS, but they went for Scality instead.
Maybe you should check out Scality's Ring technology too. ®
MPAA to send take down notice to all these Distributed hash table P2p for profit any day now ;)
MPAA and others , perhaps Antipiratbyrån (Anti-Piracy Bureau, APB) in the EU OC, to send take down notice's to all these Distributed hash table P2p for commercial profit companies any day now ;)
after all Distributed hash table and P2P are illegal the world over or so they would have you believe so Distributed hash table P2P coming to the cloud can only mean Piracy right, or that will be how they spin it if it suits their everything's piracy because we say so agenda
Not exactly new
This same sort of DHT or consistent-hashing approach is already widely used. Amazon's Dynamo and Berkeley's OceanStore both work essentially this way for different kinds of data, and have been around for over ten years. Atmos, which you even mention, inherits some of the same ideas from OceanStore. Tahoe-LAFS and GlusterFS both embody those ideas for files, Voldemort/Cassandra/Riak for key/value stores, etc. Scality might indeed have some very good technology, but the DHT part isn't their differentiating feature.
Testing 10,000 nodes
They could run 10 VM instances per physical node. Of course that'll introduce inefficiency, but it should be just a constant VM overhead, and will let them see if anything breaks down or scales badly.