Juniper Networks WXC Project

Hey all,

sorry that i didn't post a lot the last month, but i was traveling for work and had some holiday.

Some post ago i talked about a project with WXC, now i want to show you some figures about the preformance of these devices in real network.

This is from a office to hq in the same country (so no latency problems).









This image clearly show that they don't use object caching, but Molecular Sequence Reduction™ (MSR™) compression and Network Sequence Caching technologies. the first one is in memory, the second on harddisks (MSR can be done by WX, but only NSC can be done by WXC)

why do i say you clearly see it doesn't use object caching? Because when you see the image you see that at 10/07/08 someone is asking a file with a size of +/- 4.5 gig and is compressed by almost 95% (so out of cache), this can also be done by object caching, BUT then someone needs to ask the same file before (which will need to pass the line 1 time compleetly). This is not the case here (you see clearly see in the image that never before there has been a request for a file larger then 500MB.).

Some other figures:

Compression Summary

Peak Data Compression	98,00%
Total Data Compression	86,10%
Total Bytes Into Compression	17,7 GB
Total Bytes Compressed	15,3 GB
Total Bytes Out of Compression	2,5 GB
Effective WAN Capacity	7,20 X

Wow, isn't this nice ! this means that from the 17,7 GB that is asked to pass the line only 2,5 GB Really crossed the line. and i can even say that from that 2,5 GB only 1.4 GB needed to pass the line, because 1.1 GB was bypassed because of license issues! (see below)

Passthrough Data
Category Bytes Packets
No Remote WX 79.253.312
By Filter Rule 675.185.052
Overflow 68.746.046
No Local Reducer 8.299
Non-IP 210.496
License Violation 1.135.001.661

But hey which traffic includes this figures?



Now you have a idea about which traffic it is. OK, i need to be clear about one thing, this is the best figure i have of all my sites :). BUT, still it is real traffic.

Comparing WAN Accelleration Products (Juniper WXC vs Riverbed vs Bluecoat)

Hey All,

For some weeks now i am busy with a project that includes Juniper Firewalls (with VPN's), Bluecoat (WAN ACC), Juniper WXC (WAN ACC), Juniper NSM (Management station), Juniper CMS (WAN Management station).

The Firewall/VPN part was difficult, because at HQ they had 2 ISP, but also at the remote side they had 2 ISP. The customer asked full redundancy (so we needed to create 4 VPN tunnel from remote to HQ). This i did before so wasn't to hard! (just play around with Vrouters and everything is up and running).

The real hard part was which WAN Accellerator will they take! So the customer start testing together with me.

Products we tested:

Bluecoat
Juniper WXC
Riverbed

We provider the tests for Bluecoat and Juniper, After these test we saw that on a satelite link bluecoat wasn't doing well (especially when using http!)

Personal i had the feeling that bluecoats proxy kicked in here and didn't do alot of WAN Accell. We also called bluecoat and they came over to do some tweaking, and saw that the customer his webpages wheren't cacheable (and this is the reason it didn't work that well)

After the bluecoat test, they tested Riverbed (which wasn't that bad, and better then Bluecoat).
The last test we came again with Juniper WXC and it seems that this was the fastes solution.

So because of the outstanding Juniper FW/VPN concept (vrouter and routing based vpn's) and the good performance of Juniper WXC, the customer choosed us to build there international Network.