Increasing bandwidth is not always the answer

Then you get the devs that test a new application on the LAN or a single instance on the wan and it runs great and company rolls it out without consulting the networking dudes only to find out it does not work so great over the wan. Then they blame the network and the bosses want to know why the network is so crap *sigh*
 
We've installed many WAN optimsation appliances and the results are amazing to say the least.

I'm not going to mention brands or anything that may be regarded as advertising, but if any IT admins out there are interested in the concep, feel free to drop me a PM.
 
Whilst I understand the need for application optimisation, what passes for bandwidth in this country is beyond laughable.

When you process 3GB of incoming EMail daily, having a 1mb dediated circuit won't cut it, there is simply no way to manage that kind of traffic other than increasing bandwidth... remember ye ol' dutch packet, everyone ones first priority with maximum bandwidth at no cost.

I'm all for application management and do so to the best of my ability on many of our links using compression, shaping, policing, filtering on routers with plenty of other optimisation at the actual application level with remote caching and the like. in general it's not too difficult to manage but the total traffic growth has been astonishing.

In a little over a year use on a single site has grown from 30 to over 100GB monthly for neccessary web traffic, double that figure to include EMail and then add all the other remote applications using bandwidth, VoIP, VPN, AD, Exchange, DNS, FileSharing, SQL, OSPF, RIP, FTP and dare I say it, all the torrents...

you very quickly realise that this country has budgetary constraints (first and foremost) that prevent us getting the bandwidth needed to optimise application performance in the first place.

D
 
I will mention brands. :) riverbed steelhead rocks. All we have been using for 3+ years in a 100,000 workstation worldwide enterprise company. Never looked back!
 
I've been involved in several trouble-shooting scenarios where application performance issues were creating severe impact on the business - dropped transactions, SLA violations and penalties. Whilst bandwidth was the root cause in a small percentage of instances application architecture was by far the biggest contributor to poor performance. WAN optimisation is certainly a viable approach to addressing these performance issues but not in all cases. I have seen an organisation invest 10's of millions in Riverbed to try and improve performance of an application deployed across 700+ sites country-wide with no success. They then moved to CITRIX as a delivery mechanism for the application and again had no luck. Subsequent investigation revealed that the application was at fault (i.e. select * from database every time a user ran a search) with huge volumes of unnecessary data being transferred. Fixing a single SQL query had 90% impact on improving the performance of a critical transaction.

Ultimately the right tool should be used for the right purpose. I feel sorry for the NW admins as they are almost always the first scapegoats when applications do not perform. My recommendation is that before you even spend money in more bandwidth or WAN optimisation determine if that is the actual problem you need to address. There's nothing worse than throwing thousands/millions of rands at network capacity/optimisation when there is no subsequent pay-off.
 
Hi Slave.

I'm not aware of a 700 site deployment in the country - ever. Are you sure it was Riverbed?

Are you able to PM this company name?
 
Sorry for the delay - just arrived back in SA. The site I mentioned has not got riverbed deployed across all sites, only a subset.
 
Citrix has a Virtual Appliance Branch Repeater VMX, formally WanScaler, which is free for small installations/testing. Have done some testing with it and I have been very impressed with the results....
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X