The unit will now monitor traffic flow and will optimize based on available bandwidth or on the type of data. For example if bandwidth is constrained they will spend more time compressing and deduplicating before sending the information across the network. Conversely, if you have a big pipe it will do a more simple optimization path to move data across the wire going to a compression only mode. This optimization can be done on a connection by connection analysis.


Also unlike backup dedupe, Riverbed has a greater chance of already having seen the file. For example a PowerPoint file may have been emailed as an attachment and stored to a NAS share long before it is encounter by a backup replication job. In this case the WAN only has to move the data once even though it has been asked to send it at least three times.


On the future’s front, Riverbed is looking to move its data understanding beyond where it is today. They have built optimizers already for TCIP, CIFS, NFS and MAPI to improve WAN performance. They are also looking to move to supporting a more basic block protocol. Imagine being able to boot a virtual machine over the WAN? That will make consolidation very interesting... More to come on this initiative.

George Crump, Senior Analyst

SNW Briefing

   

    Additional SNW Meetings


  Spectra Logic

  Supermicro

  Avere

  Micron

  Bocada

  Ocarina Networks

  Data Robotics

  Nasuni

  FCIA

  Sepaton

  CommVault

  NextIO

  SNIA

     

Storage Switzerland is at SNW in force again this year and as in years past we will be doing our best to bring you continuous updates from the event. These reports are quick summaries of our meetings at the show; look for more detailed analysis on our blogs on Information Week and Network Computing.