Blocks and Files Generally, it seems to me that object storage is suffering from a failure to launch despite more than a dozen suppliers pushing it.
Many of these same vendors seem to have their heads in the sand with regard to their place in the marketplace - they seem to ignore the fact that end-user buyers are confused about what object-storage really is, how it gets purchased (and why) and how it gets used.
Here's a quick and brief run-down of vendors and their object storage activities:
- Amazon - biggest object storage user?
- Amplidata - object startup which has landed Quantum OEM deal
- Caringo - object storage startup which landed and then three years later lost a Dell OEM contract
- Cleversafe - talks massive scalability with a 10EB store and 1TB/sec ingest rate
- Dell - cancels Dx6000 object storage product and makes Caringo software available as a download
- DDN - says its WOS can be cheaper than Amazon storage
- EMC - Centera --> Atmos --> Project Bourne and HDFS - something big stirring
- Exablox - startup whose object storage underpins CIFS/SMB access protocol
- HDS - object is a key part of its strategy and is in the mainstream of HDS development
- HP - has its own object technology
- IBM - nothing, zip, nada except maybe vague hints of GPFS having object connections
- NEC - has Hydrastore technology using erasure coding and focussed on deduped backup
- NetApp - StorageGRID but no visible activity
- Quantum - OEMing Amplidata as its Lattus object store in StorNext and as alternative to tape
- Scality - object startup that's landed SGI and Penguin as OEMs and expands its technology's relationships energetically
- SGI - offers Scality RING as object store inside its gateway file location virtualising appliance
Quantum with StorNext and SGI with its gateway have a media and entertainment and oil and gas market presence respectively.
What is the function of object storage? Variously:
1. Alternative to tape for archive
2. Alternative to file systems for very large populations of files
3. Alternative to RAID-based data protection schemes offering greater storage efficiency and lower rebuild times
4. Aid to dedupe via object storage algorithms and hence better storage efficiency
Disadvantages and problems
1. No industry-standard object interface equivalent to, say, CIFS or NFS with files
2. No agreed storage role and not enough storage pain being felt by users/vendors to drive its adoption and development more strongly
3. Complex technology that can be hard to understand
4. No support/commitment from IBM
5. Seemingly lackadaisical support from NetApp
6. Disengagement from Dell
There could usefully be an industry Object Storage Computing Adoption Resource (OSCAR!) to provide general support and encouragement for object adoption, kind of like the Active Archive Alliance. Vendors seem to agree something more can and should be done - yet they are not doing anything about it.
The general idea I have is that object storage generally is suffering from a failure to launch despite good technology because users and mainstream vendors are not feeling the pain of problems from filesystem technologies or tape enough to do anything substantial about it.
There is pain, surely, and it needs to be surfaced and made visible.
For object storage to generally prosper we need clearly-defined object storage use cases where filesystem technology is obviously not suitable, even if that means a filesystem access protocol is used layered on top of a backend object store, and where tape is clearly inferior for a users' needs.
Come on object storage industry; you have common problems and need to get together and sort them out. ®