WAN optimization reviews

[Cooky]

Does anyone have good review references for the top WAN op vendors?
Network Computing has a nice review article, that compares Citrix WANScaler, Bluecoat ProxySG, and Cisco WAAS, which is exactly what I'm looking for, except it's 4 years old.

http://www.networkcomputing.com/data-networking-management/229608058?pgno=1

If anyone has something similar, but recent, could you please post the URL?
I'd also appreciate if you could share your experience w/ existing solution you have.

We're considering Citrix NetScaler, Bluecoat ProxySG, Cisco WAAS, and Riverbed Steelhead.

thanks.

 

[ViviTheMage]

We went through this about 8-12 months ago, we went with Riverbed. Lots of things WAAS didn't have, bluecoat wasn't even a contender.

We ended up with a 5050 as our server side appliance, 1050 in a large branch, 2050 at our DR branch, and a bunch of 250's and 550's at other branches.

We're actually running RSP + Win server 2008 on them as well...purely for print server, and eventually some sort of file storage IF we need it.

My role in this is to research/deploy/maintain.

If you get more then 15 appliances, get a CMC, you'll be happy. You'll also have pretty graphs that execs can see that will let them know it's worth the cost.

 

[spidey07]

2nd Riverbed.

 

[ViviTheMage]

This is an example pretty graph from our network after 2 months of usage (datastore's were all 100% at branches finally).

What are you trying to accomplish? What do you want optimized? How many branches, what connections/latencies?

 

[Emulex]

whats the freeware solution? it's just a big dedupe/solid zip system?

 

[ViviTheMage]

I don't think there is a good freeware version of this type of software...

 

[Cooky]

whats the freeware solution? it's just a big dedupe/solid zip system?

Only thing close is Squid, which can cache web contents, but I don't think it provides nearly as many features as the commercial solutions.

http://www.squid-cache.org/

 

[ViviTheMage]

I also played with squid, it's definitely no where near where Riverbed, or WAAS is even.

 

[Cooky]

What features did the Bluecoat ProxySG lack that made you go w/ Riverbed?

 

[w0ss]

We tested Riverbed, Silverpeak, and Cisco. We went with Cisco WAAS. It really depends what you are trying to do. We were already a Cisco shop and a NM module sounded good to us. With the new WAASX it is even cheaper.

I have looked at a few open source products. The two I looked at were very basic and were very administrator intensive.

http://www.trafficsqueezer.org/
http://wanproxy.org/

Also be prepared for the support issues that come up. One reason we went with WAAS was the fact that the traffic is not tunneled. each TCP stream will be on the network. Most competitors will take the streams and run them over some type of tunnel where firewalls and IPS devices cannot see the traffic.

 

[Railgun]

We went with Cisco as well, though we're still in a pseudo pilot phase. Though some things due to our current PC architecture won't be optimized, but in the long term, it will be of great benefit.

 

[ViviTheMage]

We went riverbed because waas didn't support a lot of what we wanted to optimize. We also have a lot of branch to branch traffic, which waas did not support.

We also wanted to run a small/mini server for print serving/dhcp at the branch, so if the connection is down, they're not dead in the water. Waas did not have that functionality.

As others have said, look at your requirements, and look at the options.

 

[Cooky]

Thank you all for the responses.

So far we have two Riverbed and two Cisco deployments.
Anyone out there that picked Bluecoat's ProxySG or Citrix's NetScaler/Branch Repeater?

 

[w0ss]

We went riverbed because waas didn't support a lot of what we wanted to optimize. We also have a lot of branch to branch traffic, which waas did not support.

We also wanted to run a small/mini server for print serving/dhcp at the branch, so if the connection is down, they're not dead in the water. Waas did not have that functionality.

As others have said, look at your requirements, and look at the options.

Those limitations do not exist anymore. You can optimize traffic between remote sites as well as run virtual machines on the appliances. Assuming you have the right appliance and software version.

 

[Zstream]

Riverbed is great.

 

[ViviTheMage]

Those limitations do not exist anymore. You can optimize traffic between remote sites as well as run virtual machines on the appliances. Assuming you have the right appliance and software version.

Good to hear, I was told they were pretty far behind with it when I asked/researched. There were a few more things we did not like WAAS for as well, but I am not recalling.

OP, look at the options, see if you can demo, that's what we did.

 

[Cooky]

thanks.
Obviously we'll evaluate, but we don't have time to do it for all the vendors.
We're limited to 3, so need to figure out which one of the four to drop from the list.

We'll do Riverbed & Bluecoat for sure, so that leaves Cisco & Citrix.
We're a big Cisco shop, so support wise it makes sense to include them.
However, Citrix seems to be the only vendor that can optimize ICA across the WAN, so we don't want to miss out on that feature.

 

[w0ss]

I am not sure how big a company you are at so you might not be able to get the same treatment. We were able to get the vendors to come onsite and configure all the equipment up all we had to do was run our load generation test. The issue we had was our testing. We were using software to pull down static web pages on a server. All of the devices were able to turn 1gbps of traffic into about 3mbps lets just say not really a valid test. All of them wanted to do live testing which we did on our SRDF/VTL portion of testing. If there is a way to do live testing I would recommend it.

 

[m1ldslide1]

Since you're a big cisco shop I would look at them closely. The fact that you can integrate with your existing routers can save a lot of money on hardware and support, both from a capex and an opex standpoint. Plus I'd give your cisco account team a chance to talk about the new versions of WAAS - I hear that its improved tremendously over the past few months, and that it competes neck-and-neck with riverbed depending on what you're trying to accomplish.

 

[Cooky]

Our hesitation w/ Cisco is actually not about what WAAS cannot do, but rather Cisco's track record of dropping something they don't think is profitable.

We invested in their Web App Firewall & XML gateway, (through acquisition of Reactivity). It's a decent product that works well, and is easy to manage.
Cisco announced to discontinue that product line last year, because it's not a billion dollar market.(or they don't think they can make that much profit)

So we were left out dry w/ no option, but migrate to Vordel.
Application networking isn't their strongest suite, so we have our concerns.

Besides, the modules that can fit into an ISR probably aren't beefy enough to support some of our remote sites, so integration doesn't really buy us that much.
We'll still do some research though, to see if it's worth looking into.

thanks again for the input.

 

[Railgun]

Cisco's track record of dropping something they don't think is profitable.

Heh....we were in the same boat with them and the Arrowpoint/CSS platform. But I digress...

But in this case, it's seen a lot of development since we first looked at them with this four years ago. I think this falls more in line with their core business than not.

 

[Emulex]

with the cost of metro-E being so cheap , how much does this really improve? 10/10 metro-E is about $600 here (SLA) and comcast business (best effort) 100/20 is $399 here. do you find these devices boost the performance that much? what kind of latency do you get and what multiplier do you get? Would you expect to be able to push 30-40megabit out of a 20 meg pipe? just curious - bandwidth is so cheap these days

 

[Cooky]

Emulex - in which location are you getting metro-E MPLS for only $600 and through which provider?

We have ~120 remote locations throughout U.S. and can't even get ethernet hand-off in some cities.
In some cases it took Verizon 14 months to upgrade from traditional T1's to ethernet.

WAN optimization is our strategy to mitigate many issues, some are completely out of our control.
And yes, we expect to be able to improve the efficiency of the WAN bandwidth by two folds, if not three folds

When these remote locations grow and need bandwidth upgrade, it's not just on the remote end - we have to plan and upgrade capacity on the head end as well.
So the costs associated w/ buying shear bandwidth is going to be more than that of WAN op in the long run.

 

[flmnfvr]

www.fatpipeinc.com. 5 load balancing algorithms, nobody does this!!! Best product in the planet, FBI / White House uses our technology.

---

Hi,

Welcome to AnandTech Forums. I locked the thread because the last previous reply was over two years ago.

Our members refer to such posts as "necro" posts, and some of them tend to post replies ranging from sarcastic to hostile, including possible accusations of spamming, whether warranted or not.

You are welcome to continue posting on our forums. Thanks for understanding. We hope you enjoy yourself on our forums.

Harvey
Senior AnandTech Moderator/Administrator