Big Blue tapes up big data
IBM's 4TB format
IBM is announcing a raft of tape-related products, including a new tape format and better robotics for its high-end tape library. It is positioning tape as an integrated archival tier of data storage.
The new TS1140 drive and format holds 4TB of native data per cartridge – 20 per cent less than the Oracle T10000c 5TB – but transfers native data at 295.5MB/sec (my derived number, based on 2.2:1 compression and 650MB/sec max compressed throughput), significantly faster than the Oracle drive's 240MB/sec. IBM claims its TS1140 has the highest linear density of any tape drive in the industry.
It uses up to 65 per cent less electricity than Oracle's T10000c drive because it has fewer and newer components. For example, the TS1140 drive can write 32 channels at once, up from 16 with a previous drive. This means it can deliver significantly better performance with fewer tape passes and at lower tape speed.
When used in the TS3500 high-end library, the TS1140 raises the maximum compressed capacity to 2.7 exabytes – 2,700 petabytes. The maximum native capacity is 900PB with TS1140 tapes, which compares to the maximum capacity of Oracle's StreamLine 8500 library using T10000c tapes at 500PB.
The TS3500 library has a new shuttle technology for its tape cartridge moving robot. Up to 16 library frames can be inter-connected – meaning a base unit with 15 expansion frames or strings. The shuttle enables a connection from any library string to any library string without an intermediate robot; it is hands-off. It fits both LTO and TS11XX-based models of the TS3500, and all models of the High Density frames are "shuttle ready". It also works with existing and past models of frames, tape drives and media, and is reconfigurable if room constraints change.
TS3500 Shuttle Complex (IBM)
IBM says the library can have up to 2,700 drives for "massive throughput," and the shuttle enables up to 15,000 mounts per hour. The SL8500 only supports up to 640 drives, according to IBM.
It claims in general that it "has an edge over storage vendors like EMC that don't support tape... IBM's Scale-out Network Attached Storage (SONAS) system and Information Archive both use policies to automatically migrate data to the tape storage tier."
LTFS and tape virtualisation
IBM is also announcing file system access to selected IBM tape libraries with the Linear Tape File System Library Edition (LTFS LE), "invented by IBM Research to provide a simple, cost efficient way to access and manage massive archives of data and digital assets". IBM added that "LTFS clients can now more efficiently index, search, retrieve and share data stored on Generation 5 LTO tape, an open tape storage format." We don't know yet which IBM tape libraries are supported.
IBM is adding a many-to-many replication feature to TS7650 ProtecTIER deduplication products. Customers with multiple datacentres can automatically replicate backup data between them to protect data. It claims ProtecTIER can strip out up to 95 per cent of the data by removing duplicated blocks, and thus reduce network bandwidth and target storage capacity needs.
The TS7700 Virtual Tape Library family of products has doubled the number of virtual tape cartridges it contains to 2 million. It has a 900MB/sec throughput enabled by using POWER7 technology, and this is a claimed 70 per cent better than the 640MB/sec of the Oracle StorageTek Virtual Storage Manager with its 640MB/sec throughput.
IBM has poured a lot of investment into its physical and virtual tape products and is emphasising that it outperforms its Oracle competition, even though its new tape format holds 20m per cent less raw data than Oracle's T10000c. IBM is also positioning the tape library as a natural destination for archival data with policies to automatically move data to this archive tier of storage and thus integrate it with disk storage tiers. ®