In summary, Riverbed devices are deployed on either end of a connection, typically a WAN or high latency connection, the devices will peer with each other and then build up tables in what Riverbed calls SDR (scaleable data referencing).
In simple terms imagine an IP conversation across a network is comprised of packets of data, if you think of these packets as words in a conversation it makes Riverbed easy to understand.
So using this example each packet equals a "word" in a conversation so if we had a short statement which said "THE BROWN DOG" when the conversation flowed through each Steelhead it will add these 'words' to its database and give each word a unique reference.
Next time the edge device sees one of these words. instead of sending the whole word(packet) it sends the reference only across the WAN link, then the remote site Steelhead rebuilds this packet and sends it out LAN side.
What this means is common and repetitive data which runs across WAN and high latency connections can be drastically reduced, if you think of emails as an example they often contain header information and signatures/logos etc, all of this equates to many packets of data which are sent repeatedly across the WAN.
The Steelheads will do run SDR on all types of TCP traffic by default but as well as this they also have specific optimisation streams for common 'chatty' types of traffic as well which optimise these even further.
These include: CIFS (including encrypted SMB1 and SMB2), HTTP(S), Oracle forms, MAPI (Exchange), SQL, NFS, Lotus Notes, Citrix (ICA and Session reliabilty), FCIP and SRDF.
These include: CIFS (including encrypted SMB1 and SMB2), HTTP(S), Oracle forms, MAPI (Exchange), SQL, NFS, Lotus Notes, Citrix (ICA and Session reliabilty), FCIP and SRDF.
In the deployments we manage at SAS we have seen drastic reductions in areas such as print traffic (again something which sends a lot of repeat information) and especially CIFS (such as Windows mapped drives) and MAPI (Outlook to Exchange conversations) The end result gives the user a much more LAN-like experience when working either remotely or in a remote office.
For mobile workers Riverbed has a mobile client which runs on laptops and acts as a built-in accelerator. This enables remote/home working to be much faster and more productive by indirectly reducing the latency of working on Internet connections. This mobile client works in the same way as an appliance with SDR but uses the local hard drive to store its database and reference
No comments:
Post a Comment