Announcing Spectrum Scale 5.0

Late last year Spectrum Scale 5.0 was released, this is the first major version of Scale (formerly known as GPFS) for quite some time and represents some major updates to the system. This includes making day-to-day operation far easier via GUI developments and large improvements to installation.

For those who don’t know Scale is a fully software defined highly configurable, scale-out storage system, which is used in archive, data protection storage, cloud object storage, NAS, Hadoop, medical and HPC environments. You don’t need to buy hardware from IBM, as with many competitors’ systems, but IBM do produce hardware that it can be run on – don’t think of this as an appliance though, rather an assured performance platform. Unlike an appliance, once you’re finished with the hardware, you can do what you want with it, it’s not locked to the software. Scale can also be run on your own hardware, in the cloud or extended across multiple sites and cloud.

As part of the Spectrum Storage Suite, Scale is tightly integrated with the other Suite packages. For example: Scale has backup integration with Protect, however Protect can also integrate with Scale as storage. Scale can be monitored by Control, but also perform HSM to Cloud Object Storage and/or Archive…

IBM’s announcement and new features can be found here:

There’s a really large amount of new features, it’s worth a look.


Looking at Scale-Out PBBAs? Beware of hidden costs!

I’ve been doing some research into various of the new generation of Scale-Out backup products – you know the ones, they’re commodity Intel based servers with some disks and some flash cache. They’re dedicated to their backup software, not a scale-out product like IBM’s Spectrum Scale, or the various competitor general purpose scale-out systems. These appliances are sold with many promises, including:

  • Erasure coding means hardware redundancy and fast disk rebuilds
  • Ease of infrastructure expansion – just add new nodes
  • Massive Scale-out, near limitless data storage
  • Enormous bandwidth for ingest

However, upon closer inspection these are all features available from IBM in both hardware and Spectrum software defined storage.

A backup system can run on storage presented by Spectrum Scale, Accelerate or Virtualise, or use a hardware array such as Storwize v5000. All these systems can be dynamically expanded with just a few mouse clicks, all the while remaining online and delivering dRAID protected disks (or similar) for fast rebuild and hardware redundancy. These systems can scale out massively to multi PB size and even larger. Servers can be implemented with multiple teamed NICs to deliver huge bandwidth. Essentially all the “unique” features of the new generation of PBBA are covered off however, as we dive deeper we start to come across a major disadvantage of the new generation of PBBAs.

Scale-out PBBAs’ dirty little secret

Power consumption. I looked at three of the market leaders’ example configurations and it’s not pretty:

                               Usable Storage        Power Consumption
Company 1          32TB                           1.7KW (average)
Company 2          72TB                           1.99KW (max)
Company 3          205TB                        4.8KW (max)

To hold this up to a comparison with IBM, I looked at a Lenovo SR630, with 2x26core Xeons @ 2.1GHz, with 256GB RAM (ie: enough to run virtualised Spectrum Protect and Protect Plus on the same box) with storage presented from a single Storwize v5000 shelf, it looks like this:

                                                   Usable Storage   Power Consumption
V5000 SSD & SR630               245TB                    329W (average)
V5000 SAS NL & SR630         80TB                     534W (average)

IBM Data Protection competitors’ new generation scale-out PBBAs use an absolute minimum of three times the power.

And we haven’t even looked at cooling yet…

Spectrum Protect Suite 8.1.4 Announcement

The Spectrum Protect Suite (not to be confused with Spectrum Storage Suite…) has been announced. New features include

  • Simple, automated, and customizable operational reporting in the operations center
  • Simplified configuration with quicker time to value for virtual environments
  • Faster recoveries when restoring multiple virtual machines for VMware
  • Ability to restore Microsoft™ SQL Server Volume Shadow Copy Service (VSS) backups to alternate instances
  • Automated creation of SQL database clones through storage snapshots
  • Expanded policy options for SAP HANA backups
  • Simplified configuration and expanded support for Oracle with Spectrum Protect Snapshot
  • Space management support for Linux® on Power® Little Endian
  • New enhancements for space management on UNIX® and Linux and Microsoft Windows™

Full details can be found here: