The Riverbed Blog (testing)

A blog in search of a tagline

Archive for May, 2011

Last week’s Packet Capture Innovators event

Posted by riverbedtest on May 31, 2011

Today, we're featuring something a little different.  What follows is the transcript of the Packet Capture Innovators Twitter Party/Interview that we did last Tuesday, May 24.  Thank you to everyone who has asked us to post it, and to everyone who took part.

LauraChappell
Getting ready to check out Twitter interview with Gerald Combs – creator of #Wireshark! Filter #packetcap to follow!
5/24/11 11:50 AM

Riverbed
Pull up a chair; we'll be starting the #packetcap event in another 5 minutes…
5/24/11 11:55 AM

Riverbed
Welcome to #Riverbed's first-ever Twitter Party! We'll hear the story of Packet Capture from the guys who created it! #packetcap
5/24/11 12:00 PM

Riverbed
That'd be #Riverbed CTO & co-founder Steve McCanne (SM) & CACE Technologies founders Loris Degioanni (LD) & Gerald Combs (GC). #packetcap
5/24/11 12:00 PM

Riverbed
We're calling them (and the event) The Packet Capture Innovators. #packetcap
5/24/11 12:01 PM

Riverbed
Steve created #tcpdump & #libpcap back in 1986, Loris wrote the Windows versions #WinDump & #WinPcap in late the 90s. #packetcap
5/24/11 12:01 PM

LauraChappell
Loris Degioanni – creator of #WinPcap on Twitter interview – filter on #packetcap!
5/24/11 12:02 PM

Riverbed
And Gerald wrote #Wireshark (originally called #Ethereal) in 1998. #packetcap
5/24/11 12:02 PM

Riverbed
Today we're going to tweet their story and they'll field questions from everyone who is following us. #packetcap
5/24/11 12:02 PM

Riverbed
So start sending your questions for Steve, Loris & Gerald with our hashtag #packetcap, and we'll answer them at the end.
5/24/11 12:03 PM

Riverbed
RT it and tell your friends! And be sure to use our hashtag #packetcap! And please follow @riverbed
5/24/11 12:03 PM

Riverbed
Let's get this party started! #packetcap
5/24/11 12:04 PM

Riverbed
Steve McCanne: In 1988 I was an undergrad at UCBerkeley & took a summer job with Van Jacobson at Lawrence #Berkeley Labs. #packetcap
5/24/11 12:04 PM

Riverbed
SM: Van had been working on the congestion collapse problems in the #ARPANET and added congestion ctrl to TCP. #packetcap
5/24/11 12:05 PM

Riverbed
SM: Traffic analysis overwhelmed the LAN and system doing the watching. I suggested a filtering model to limit the traffic flow #packetcap
5/24/11 12:05 PM

Riverbed
Remember to send your questions for Steve, Loris and Gerald to us at @riverbed with the hashtag #PacketCap! We'll answer them at the end
5/24/11 12:06 PM

Riverbed
SM: The filter lets user pick the traffic to look at. The kernel driver would keep what user wants, and junk the rest. #packetcap
5/24/11 12:06 PM

Riverbed
SM: I combined the filter with a compiler so users could filter with a high-level language. #packetcap
5/24/11 12:07 PM

Riverbed
SM: I combined the filter with a compiler so users could filter with a high-level language. #packetcap
5/24/11 12:07 PM

Riverbed
SM: We took the 2 parts, added a tool to run it, and released it to #OpenSource community as #tcpdump. #packetcap
5/24/11 12:07 PM

Riverbed
SM: Designed for #Unix-like systems. The project developed a broad following right away. #packetcap
5/24/11 12:08 PM

FOSSwiki
SM: We took the 2 parts, added a tool to run it, and released it to #OpenSource community as #tcpdump. #packetcap http://bit.ly/kjEEbs
5/24/11 12:08 PM

Riverbed
SM: I wanted to write other packet capture apps, so we turned #tcpdump components into #LibPcap, a library & the pcap file format #packetcap
5/24/11 12:08 PM

Riverbed
SM: It became the de facto standard for building packet capture and network analysis apps. But it was hard to use… #packetcap
5/24/11 12:09 PM

Riverbed
Remember to send your questions for Steve, Loris and Gerald to us at @riverbed with the hashtag #PacketCap! We'll answer them at the end!
5/24/11 12:10 PM

vmlemon
Wondering if the WinPCap folks plan to migrate to NDIS6, so that Windows Vista/7 users can finally capture 802.11 traffic. #packetcap
5/24/11 12:10 PM

Riverbed
Loris Degioanni: Late 90s I was a student at Politecnico di Torino in Turin, Italy. #packetcap
5/24/11 12:10 PM

Riverbed
LG: Assigned a project to build a network analyzer for students to use in #Windows. #packetcap
5/24/11 12:10 PM

Riverbed
LG: I didn't know where to start. Some online research led me to a paper by SM and Van. #packetcap
5/24/11 12:11 PM

Riverbed
LG: The paper let me port #libpcap to Windows (#WinPcap) and I followed their lead and made it available in #OpenSource #packetcap
5/24/11 12:11 PM

Riverbed
LG: Later I ported #tcpdump to Windows just to see if it would work. It did. #packetcap
5/24/11 12:12 PM

Riverbed
LG: We put it on the school's web site, and it got tens of thousands of downloads. But it was not user-friendly… #packetcap
5/24/11 12:12 PM

Riverbed
Gerald Combs: I was working for an ISP in Kansas City using #tcpdump and #snoop all the time. #packetcap
5/24/11 12:13 PM

Riverbed
GC: But I needed something more interactive that would let me browse packet traffic. #packetcap
5/24/11 12:13 PM

Riverbed
GC: So I started working on a weekend project that, um, got out of hand. In 1998 I released the first version of Ethereal. #packetcap
5/24/11 12:14 PM

Riverbed
GC: Ethereal was #Linux and #Solaris only. I had no interest in writing #Windows software at the time. #packetcap
5/24/11 12:14 PM

Riverbed
GC: #Windows is by far our most popular platform. #WinPcap has been essential to our success. #packetcap
5/24/11 12:15 PM

Riverbed
Remember to send your questions for Steve, Loris and Gerald to us at @riverbed with the hashtag #packetcap! We'll answer them at the end
5/24/11 12:15 PM

Riverbed
GC: We renamed #Ethereal to #Wireshark in May 2006 when I joined CACE. #packetcap
5/24/11 12:16 PM

Riverbed
GC: My previous employer owned the #Ethereal name and when I left, we had to change it. #packetcap
5/24/11 12:16 PM

Riverbed
GC: One of the names we considered was Ether Weasel. But that was overruled. #packetcap
5/24/11 12:17 PM

Riverbed
GC: In the end, #Wireshark was chosen by our Official Style Consultant (my wife). #packetcap
5/24/11 12:17 PM

LauraChappell
@riverbed THANK GOODNESS! Ether Weasel sounds like a porn project! #packetcap
5/24/11 12:18 PM

Riverbed
GC: 2 communities (Users and Developers) quickly formed, and they've continued to grow to this day. #packetcap
5/24/11 12:18 PM

Riverbed
GC: In 2006, I realized that people needed to capture and analyze wireless packets, something that #Wireshark was not good at. #packetcap
5/24/11 12:18 PM

tedsseveredhead
Loris, how has the WinPCap library changed from the early days to now? #packetcap
5/24/11 12:19 PM

dances_w_vowels
Steve McCanne, what was the first computer you ran TCPDUMP on? #packetcap
5/24/11 12:19 PM

Riverbed
GC: So I reached out to Loris and we teamed up. #packetcap
5/24/11 12:19 PM

Riverbed
Remember to send your questions for Steve, Loris and Gerald to us at @riverbed with the hashtag #packetcap! We'll answer them at the end!!
5/24/11 12:19 PM

Iceklube
Can you tell us what the future looks like for #WinPcap? And do you have a few spare Airpcap Nx USB adapters you can send my way #packetcap
5/24/11 12:20 PM

Riverbed
LG: And that's basically how #CACE Technologies started. We did consulting for a year, and then moved into products #packetcap
5/24/11 12:20 PM

Riverbed
LG: One of our first products was #AirPcap, a wireless version of the #WinPcap library #packetcap
5/24/11 12:20 PM

LauraChappell
#packetcap Do you have a Sharkfest Discount Code for twinterview followers?
5/24/11 12:21 PM

Riverbed
LG: After a few years, companies came sniffing around looking to buy us. #packetcap
5/24/11 12:21 PM

Riverbed
LG: We weren't that interested. Business was good and we were having fun. #packetcap
5/24/11 12:21 PM

LauraChappell
#packetcap Bet folks want to know about #Wireshark on an iPhone/iPad…
5/24/11 12:22 PM

Riverbed
LG: #Riverbed called and we met with them. We were very impressed: lots of smart people and a clear vision. #packetcap
5/24/11 12:22 PM

INSPIRIT_BRASIL
PERFOMANCE DE TI: LG: After a few years, companies came sniffing around looking to buy us. #packetcap
5/24/11 12:22 PM

Riverbed
LG: In the 1st meeting (May 2010) there were all engineers, and one guy sitting quietly in the corner. Named Steve McCanne. #packetcap
5/24/11 12:22 PM

adownie
Loris, What resource do you recommend to learn more about WinPCap? #packetcap
5/24/11 12:23 PM

kimogoree
Gerald, What is your favorite Wireshark book? #packetcap
5/24/11 12:23 PM

Riverbed
LG: I didn't realize it at first, but it was the same Steve McCanne! We began exchanging notes and emails. #packetcap
5/24/11 12:23 PM

RobWilen
Loris, now that @Riverbed is sponsoring #Wireshark, are there plans to charge for it? #packetcap
5/24/11 12:23 PM

Riverbed
LG: And in the end, #Riverbed acquired #CACE in October 2010. #packetcap
5/24/11 12:23 PM

drapps
RT @LauraChappell: #packetcap Bet folks want to know about #Wireshark on an iPhone/iPad… |WHOA
5/24/11 12:24 PM

Riverbed
So the story ends happily. So between the #Cascade team, Steve, Loris, and Gerald, #Riverbed really does know packet capture. #packetcap
5/24/11 12:24 PM

Riverbed
That brings us to the end of the story. Now it's time to take questions from the audience. #packetcap
5/24/11 12:24 PM

BillRingle
Loris, what is the difference between #WinPcap and #WinPcap pro? #packetcap
5/24/11 12:25 PM

darylgoard
RT @LauraChappell: #packetcap Bet folks want to know about #Wireshark on an iPhone/iPad. \\YES YES!
5/24/11 12:25 PM

hfriedma
Loris, can you provide a peek at what the future holds for #WinPcap? #packetcap
5/24/11 12:25 PM

jamesjmoore
@riverbed "sniffing" around, you say? 🙂 #packetcap
5/24/11 12:25 PM

revolutionwifi
RT @laurachappell: #packetcap Bet folks want to know about #Wireshark on an iPhone/iPad… –> YES!
5/24/11 12:26 PM

Riverbed
@jamesjmoore Yes. We choose our words carefully! 🙂 #packetcap
5/24/11 12:26 PM

Riverbed
@vmlemon LD: We are not aware of any active development in this area. Would love to work with the #packetcap community.
5/24/11 12:26 PM

rrognlie
Gerald, was the original Wireshark really developed during a summer? #packetcap
5/24/11 12:27 PM

Riverbed
@tedsseveredhead LG: More OS support. dropped Win95 support #packetcap
5/24/11 12:27 PM

Riverbed
@tedsseveredhead LG: More OS support. dropped Win95 support #packetcap
5/24/11 12:27 PM

Riverbed
@dances_w_vowels SM: Sun 3/50 #packetcap
5/24/11 12:27 PM

jamesjmoore
@riverbed you did. I'm and expert on Expert and not on Wireshark. Do I want to learn yours? Laura? #packetcap
5/24/11 12:27 PM

dhmspector
Steve, What was your biggest take-away from your #TCPDUMP dev experience? #packetcap
5/24/11 12:27 PM

Riverbed
@iceklube LG: Performance enhancements and more platforms #packetcap
5/24/11 12:28 PM

Riverbed
@LauraChappell LG: discount code to be tweeted this afternoon #packetcap
5/24/11 12:28 PM

Riverbed
@BillRingle: LG: Pro is commercial and embeddable #packetcap
5/24/11 12:29 PM

netscantools
WinPcap is an integral part of NetScanTools Pro 11.02. Thanks Loris! -Kirk #packetcap
5/24/11 12:29 PM

fillmoremillard
What do you see as the next big thing for Wireshark? #packetcap
5/24/11 12:30 PM

LauraChappell
Live Twinterview going on with creators of Wireshark, WinPcap, tcpdump – use #packetcap
5/24/11 12:30 PM

Riverbed
@rrognlie GC: After prep, coding was done over a long summer #packetcap
5/24/11 12:31 PM

Riverbed
@dhmspector SM: Its easy to make things hard. Its hard to make things easy. #packetcap
5/24/11 12:31 PM

cloyd101
steve, what was your biggest challenge developing #tcpdump? #packetcap
5/24/11 12:32 PM

Riverbed
@fillmoremillard GC: More intelligence in capture data, annotation, etc. #packetcap
5/24/11 12:33 PM

aragonesb
Steve, How long did it take you to develop your first working TCPDUMP application? #packetcap
5/24/11 12:33 PM

Riverbed
@aragonesb SM: A couple of months #packetcap
5/24/11 12:34 PM

Riverbed
@aragonesb SM: A couple of months #packetcap
5/24/11 12:34 PM

Riverbed
Don’t forget: SHARKFEST is June 13-16 at Stanford U. Details are at http://www.sharkfest.org #packetcap
5/24/11 12:35 PM

Riverbed
#packetcap $100 discount for #SHARKFEST by using "SPONSREF"
5/24/11 12:36 PM

Riverbed
Thank you for joining us; more answers are coming. While you're waiting, please remember to follow @riverbed #packetcap
5/24/11 12:36 PM

revolutionwifi
I love Wi-Fi Pilot. What is the roadmap for continuing to expand wireless capabilities? @riverbed #packetcap
5/24/11 12:36 PM

wiredwiz
@darylgoard I second that, I would love to be able to run wireshark on my ipad #packetcap #Wireshark
5/24/11 12:37 PM

Riverbed
@MutTuja #packetcap You should be able to export to pcap then run wireshark/tcpdump
5/24/11 12:38 PM

yesthattom
Gerald, what devices can #Wireshark use to capture packets? #packetcap
5/24/11 12:38 PM

Riverbed
Keep those questions coming; Steve Loris and Gerald are here to answer them live. Use the #packetcap hashtag
5/24/11 12:40 PM

Riverbed
@yesthattom GC: Anything running winpcap or libpcap #packetcap. Obviously Riverbed products
5/24/11 12:40 PM

netscantools
Any chance WInPcap Pro can be made *not* to require admin privs? #packetcap
5/24/11 12:41 PM

Riverbed
@netscantools #packetcap Windows requires admin privilege to install driver
5/24/11 12:42 PM

netscantools
What about WinPcap support for sending packets out of things like cell modems? #packetcap
5/24/11 12:42 PM

JackKnoff
#packetcap Who's bright idea was it to set this up as a twitter event?
5/24/11 12:43 PM

Riverbed
@netscantools #packetcap OS limitation requiring a longer answer than what Twitter provides
5/24/11 12:44 PM

netscantools
yup, I know the mechanism used to installing driver, just wondering if there was any change. thx. Kirk #packetcap
5/24/11 12:44 PM

netscantools
Loris can send me a >140 char email about OS limit and register his S/W while he's at it. #packetcap 🙂
5/24/11 12:46 PM

Riverbed
@netscantools #packetcap LG: Lets chat at #SHARKFEST
5/24/11 12:47 PM

netscantools
OK – sure – just 3 weeks away! I'm ready. #packetcap
5/24/11 12:48 PM

rvbdusergroup
Ask packet capture innovators questions live by using hashtag #packetcap on twitter
5/24/11 12:49 PM

Riverbed
Don't forget to follow @Riverbed #packetcap
5/24/11 12:49 PM

Riverbed
@wiredwiz #packetcap #Wireshark iPad support would be great. Apple makes it hard.
5/24/11 12:52 PM

Riverbed
Keep those questions coming; Steve Loris and Gerald are here to answer them live. Use the #packetcap hashtag.
5/24/11 12:53 PM

Riverbed
Thank you to everyone who joined us for the Packet Capture Innovators event. #packetcap
5/24/11 12:58 PM

Riverbed
Special thanks, of course, to our special guests: Steve McCanne, Gerald Combs, and Loris Degioanni. #packetcap
5/24/11 12:59 PM

netscantools
I have too many questions for 140 chars. Thanks for doing this. I didn't get to meet Steve in Bellevue, maybe at Sharkfest. #packetcap
5/24/11 12:59 PM

Riverbed
Please follow us at @riverbed and come to #SHARKFEST June 13-16. #packetcap
5/24/11 1:00 PM

Riverbed
And check out our blog at http://blogs.riverbed.com. #packetcap
5/24/11 1:00 PM

Riverbed
Thank you and goodbye. #packetcap
5/24/11 1:00 PM

LauraChappell
Thanks folks! #packetcap
5/24/11 1:14 PM

vaxen_var
RT @Riverbed: And check out our blog at http://bit.ly/mNRTdR/ #packetcap
5/24/11 1:28 PM

da_BiGKahuna
RT @LauraChappell: Getting ready to check out Twitter interview with Gerald Combs – creator of #Wireshark! Filter #packetcap to follow!
5/24/11 1:33 PM

da_BiGKahuna
RT @LauraChappell: #packetcap Bet folks want to know about #Wireshark on an iPhone/iPad… <- Gotta check dis out 😉
5/24/11 1:35 PM

da_BiGKahuna
RT @netscantools: WinPcap is an integral part of NetScanTools Pro 11.02. Thanks Loris! -Kirk #packetcap
5/24/11 1:35 PM

netcooper10
Steve, as co-founder and CTO of #Riverbed, do you still get to write code? Do you still enjoy it? #packetcap
5/24/11 1:37 PM

cloudshark
Want to view your capture files on an iPad or iPhone? Check us out! http://t.co/76UkKOD #packetcap
5/24/11 2:24 PM

wiredwiz
@riverbed That should almost be a programming slogan with #Apple, "Apple makes it hard" #packetcap #Wireshark
5/24/11 2:31 PM

MatrixSCIT
cloudshark: Want to view your capture files on an iPad or iPhone? Check us out! http://t.co/76UkKOD #packetcap: … http://bit.ly/mympQ8
5/24/11 2:43 PM

Riverbed
@netcooper10 #packetcap Occasionally I write small, experimental prototypes, but I wish I had more time to work on larger coding projects.
5/24/11 2:52 PM

wescotttech
riverbed: @netcooper10 #packetcap Occasionally I write small, experimental prototypes, but I wish I had more tim… http://bit.ly/jm4ZmA
5/24/11 3:43 PM

TCinstallations
RT @LauraChappell: #packetcap Bet folks want to know about #Wireshark on an iPhone/iPad…
5/24/11 3:51 PM

tnagareshwar
I will be shortly doing private Q/A with creators of #Wireshark, #Tcpdump, #Winpcap, Do send your questions #packetcap http://bit.ly/mNPuuk
5/25/11 3:28 AM

securityxploded
I will be shortly doing private Q/A with creators of #Wireshark, #Tcpdump, #Winpcap, Do send your questions #packetcap http://bit.ly/mNPuuk
5/25/11 3:28 AM

Posted in Uncategorized | 1 Comment »

Boosting the cloud with Riverbed and Akamai

Posted by riverbedtest on May 27, 2011

At Interop earlier this month, Bob Gilbert, our technology evangelist, briefly interviewed me at our booth in the exposition hall. He asked what I would be exploring during one of the panel discussions I joined; I mentioned “cloud bouncing” and our recent announcement with Akamai. I’d like to explore that idea a bit more today.

A more accurate term would be “cloud boosting.” As organizations come to rely on SaaS applications for daily use, they’re beginning to realize the need for ensuring consistent user experiences regardless of the physical locations of clients and servers. However, many business applications exhibit characteristics that could undermine the user experience when distance is involved:

  • They’re highly interactive between the client and the server
  • They rely on chatty and inefficient application-level protocols
  • They move massive amounts of data between locations; this data is often redundant

Firewatch In these and other cases, WAN optimization technology in the cloud improves the user experience and increases productivity in exactly the same ways it does for traditional branch office-to-data center deployments. And for applications that aren’t sensitive about placement but fail miserably over high-latency links, WAN optimization is the key to moving those off-premise: additional bandwidth won’t eliminate the problems of long round trip times.

Classic WAN optimization places technology at both ends of the network connection—close to the applications in the cloud and close to the users in branch offices or on mobile clients. Alternative asymmetric optimization technologies, which require deployment at only one end of a network connection, might seem more attractive. However, they can’t offer the same benefits that symmetric optimization provides.

Consider the case of optimizing only the client side. Such devices act primarily as content caches and can’t help reduce the amount of data flowing between the cloud and multiple branch offices. When a user in one branch modifies a document, the optimizer updates both its cache and the original in the cloud. When a user in a another branch requests the now modified document, the entire file is sent to that branch—the optimizer there has no way to determine what was updated and the application can’t ask the server in the cloud to deliver just the changed portion. Client-side asymmetric optimization doesn’t work well in point-to-multipoint deployments, which are actually very common deployment scenarios.

Or consider the case of optimizing only the cloud side. All data transfers originating in a branch office or from a mobile user and going to the cloud will be the full amount, with no optimization at all. Data coming from the cloud might be optimized somewhat but consider this scenario:

  1. You request a 100 MB file; it’s compressed to 30 MB
  2. You make a very small edit
  3. You return the file, all 100 MB of it

That’s a 130 MB exchange in a situation which the client’s upload speed is likely only a fraction of its download speed. A much better scenario involves symmetric optimization:

  1. You request a 100 MB file; it’s compressed to 30 MB
  2. You make a very small edit
  3. You return the file; the mobile or branch office appliance detects the changed portion and uploads only that

What once was a bulky exchange in mostly the wrong direction is now just a small number of bytes sent over the WAN. Only with symmetric optimization can you accelerate to LAN-like performance.

So how can you take full advantage of symmetric optimization if you’re using applications at a provider that hasn’t deployed Riverbed in its cloud? Through our partnership with Akamai. Apurva Davé recently wrote about our joint announcement. Let’s examine it a bit more.

Akamai will place Cloud Steelheads in their hundreds of POPs around the globe. It’s likely that one of those POPs is topologically near your cloud provider, which will allow you to create optimized connections that terminate close to your ultimate destination. Riverbed will incorporate Akamai's Internet optimization technology, which provides a better route through the Internet. All you need to do is enable the Akamai service on your Steelheads; from there, we'll take over and figure out the best route to the entry point of Akamai's cloud. With their broad Internet reach and numerous peering agreements, Akamai will deliver optimized traffic over the WAN and bring it close to your SaaS application in the cloud. Optimized traffic will transit the long distance portion of the connection; unoptimized traffic will transit only the short distance to the cloud application.

You have a couple options for configuring your side of the connection. One would be to route all SaaS application traffic from branch offices to your primary data center; then it would move to Akamai’s network for delivery to your provider. Another option would be to configure each of your branch Steelheads to route SaaS application traffic directly to Akamai for delivery, bypassing your data center. If you’re using a hybrid application, with some data in your data center and some in the cloud, you’d use the former option. If the application is entirely in the cloud, you could use either option.

Rvbdakam Riverbed and Akamai enable all the benefits of symmetric optimization while allowing you to choose almost any SaaS application, no matter where you or the application reside. We’re both actively developing the technology to make this possible, and we expect to announce availability in early 2012.

Posted in Uncategorized | 1 Comment »

Packet Capture Innovators Wrap-Up

Posted by riverbedtest on May 26, 2011

Tuesday was our much-publicized Packet Capture Innovators event on Twitter.  Thanks go out to our guests of honor, Steve McCanne, Gerald Combs, and Loris Degioanni, and to everyone who attended the event, and especially to those who asked questions.  Turnout and response were very positive, especially for a first-time event like this.

vmlemon
Wondering if the WinPCap folks plan to migrate to NDIS6, so that Windows Vista/7 users can finally capture 802.11 traffic. #packetcap
5/24/11 12:10 PM
Riverbed
@vmlemon LD: We are not aware of any active development in this area. Would love to work with the #packetcap community.
5/24/11 12:26 PM

 

tedsseveredhead
Loris, how has the WinPCap library changed from the early days to now? #packetcap
5/24/11 12:19 PM
Riverbed
@tedsseveredhead LG: More OS support. dropped Win95 support #packetcap
5/24/11 12:27 PM

 

 

dances_w_vowels
Steve McCanne, what was the first computer you ran TCPDUMP on? #packetcap
5/24/11 12:19 PM

Riverbed
@dances_w_vowels SM: Sun 3/50 #packetcap
5/24/11 12:27 PM

 

dhmspector
Steve, What was your biggest take-away from your #TCPDUMP dev experience? #packetcap
5/24/11 12:27 PM
Riverbed
@dhmspector SM: Its easy to make things hard. Its hard to make things easy. #packetcap
5/24/11 12:31 PM

Thanks again to everyone who participated.  You can read the entire twitter exchange on Twitter by searching for the hashtag #packetcap.

Posted in Uncategorized | Leave a Comment »

Riverbed wins a “Larry award” for best presentation at Interop

Posted by bobegilbert on May 25, 2011

NW
  LarryAwards

Network World's blog recently featured a story dedicated the 3rd annual Larry Awards, which is a yearly contest featuring blogger Larry Chaffin's picks for best presentation at Interop.  Riverbed was selected in the network optimization category.

Here is the snippet from the blog featuring Riverbed.

Network Optimization: Riverbed Technology
Riverbed really has it together when it comes to Interop. They have a great presentation, live demos and real customers talking. The booth is always filled, everyone can answer a question and they had executives in the booth, like Arista. The only thing that I can tell Riverbed to do differently is to present Cascade on the opposite side of the main presentation. They could have one running at all times on opposite sides. I like the Cascade product and think they need to show it more. (Disclosure, I am a Riverbed reseller, but that doesn't mean I automatically like their product presentations at Interop. They still have to wow me to get a Larry Award.)

Thanks Larry for the recognition!  A tremendous amount of effort went into the planning and preparation to ensure that Riverbed had a successful showing at Interop.  I like the idea about showcasing our Cascade product more.  We will definitely take that into consideration next time. 

 

Posted in Uncategorized | Leave a Comment »

It’s official – businesses are doing less for disaster recovery

Posted by riverbedtest on May 25, 2011

At Riverbed, we spend so much time along with our partners like EMC, NetApp, HP, and others talking about disaster recovery that we assume everyone has heard us by now. I've even had those awkward cocktail party moments where I start rattling off some DR story and someone says, "Yeah, you've told me that one before." Oops.

Thankfully, the folks at Forrester are refilling my cocktail story trough with new data in their report, "State Of Enterprise Disaster Recovery Preparedness, Q2 2011."

I found two trends from this report really interesting:

  • Companies are reducing the number of DR sites they manage. It's no surprise that the recession has had something to do with this. And it's not a question of ROI (Companies noted on average $350,000 per hour as the cost of downtime), it's cold hard cash that's driving companies to fewer sites.
  • A whopping 65% of enterprises are interested or already implementing some sort of cloud/service based DR strategy. It's driven by the instant-on, elastic capacity nature of cloud and the desire to eliminate management of additional sites.

I think that these dynamics are certainly aligned with what's been driving the interest in DR-oriented performance solutions. Customers are looking to solutions like Steelhead to cut the cost of backing up or replicating data within their private environments; they look to solutions like Whitewater to accelerate and deduplicate cloud storage environments, and finally Cloud Steelhead to accelerate compute and applications in the cloud.

Customers tend to like the idea that Riverbed has a multi-pronged approach to optimizing disaster recovery, as we have to all recognize that priorities change, budgets change, but the need to protect your business doesn't.

I'd be interested in hearing from you – has your business been investing more or less in disaster recovery this year than in the past? And are you considering cloud as a DR strategy?

Posted in Disaster Recovery, Storage Cloud | 1 Comment »

THIS MORNING at 11am PT! Meet the guys behind Wireshark, WinPcap, and TCPDUMP

Posted by riverbedtest on May 23, 2011

Packet-capture-event-banner2

We've been talking it up now for a few weeks, but today is the big day.  This morning at 11am PT (2pm ET), Riverbed will host our very first Twitter party, "Meet the Packet Capture Innovators."

Steve McCanne (TCPDUMP), Gerald Combs (Wireshark, originally called Ethereal), and Loris Degioanni (WinPcap) will all be there and will share their stories with us.

In case you've missed the details, we'll start with the guys telling the story of how they started the products, and then how they met each other and eventually came to be working at Riverbed.  Once the story is done, they will be available to take questions.

So please bring your questions about Packet Capture, and remember the Twitter hashtag #packetcap so you can find us! 

You can get more details in the other blog post about the event.

We look forward to seeing you!

Posted in Uncategorized | 1 Comment »

WAN optimization in hardware?

Posted by riverbedtest on May 20, 2011

Today's guest blogger is Mark Day, Riverbed's Chief Scientist.

I’ve been hearing more questions lately about hardware approaches to WAN optimization, so I thought it would be worth looking a little at hardware/software tradeoffs – both the issues that are intrinsic to WAN optimization and some specific choices we’ve made at Riverbed. 6c_-_Compaq_386
 
Initially, Steelhead appliances and all their competitors were built as software running on conventional Intel-architecture servers.  Customers would sometimes ask about special hardware, but the only slightly-special hardware was the fail-to-wire network interface cards that allowed the appliances to be deployed in-path.  Even those NICs were off-the-shelf components, they were just a little exotic compared to the rest of the plain-vanilla server.
 
Our appliances are now custom-designed by Riverbed – we don’t typically use any hardware manufacturer’s off-the-shelf server any more. We have made a number of changes in the design to improve their reliability, based on our experiences with Steelheads being deployed all over the world.  So in a sense we do have “custom WAN optimization hardware” – but from a software perspective it’s just another x86 server.
 
Product_3 A lot of people assume that we’d get a lot more performance by using special-purpose hardware instead of a general-purpose server.  There are two major reasons why this thinking doesn’t necessarily apply.  The first, which is easier to understand, is that even though we might be able to get higher performance with special purpose hardware, it might not be economical to build that hardware.  The second reason is subtler but arguably more important – that’s due to the “moving target” nature of both our market and the capabilities of widely-used servers.
 
Let’s take up the first issue, and initially imagine that our requirements are relatively static.  In concept, it’s not hard to build a special-purpose machine that has higher performance than a general-purpose machine for WAN optimization. But as we try to sell the resulting product, we’ll generally win the contest only if that custom hardware delivers better price/performance than a commodity platform.  Just getting better top-end performance is not necessarily a win for most users. Achieving an economic win for our custom hardware may be hard, because we have to amortize the development cost of our custom hardware over a small base of buyers while the development cost of a generic server gets spread more broadly.  Even though our custom hardware may be the only available solution for certain high-end WAN optimization uses, the customers with those high-end problems might not get enough value from the solution to justify paying the price we’d have to charge. Covini-c6w-supercar_100190080_m
 
So far we’re in pretty comfortable territory.  There’s nothing very special about this first tradeoff, and you can see something similar if you look at “supercars” – people do build special high-end vehicles, but they wind up being very expensive and having a tiny market, just as we might expect for our custom WAN optimization hardware.
 
But in current-day IT there’s another factor that’s arguably more important, and doesn’t really have a direct analog when we think about supercars: Moore’s Law.  The commodity hardware is not at a single price/performance point that we have to beat just once – instead, it’s on an escalator that keeps improving its price/performance over time.  Then as if Moore’s Law didn’t add enough complexity, Plot there’s also a corollary phenomenon: the requirements for WAN optimization change over time as protocols evolve.
 
We had a fairly vivid experience with this phenomenon around our SSL optimization feature. At the point where we introduced it, we expected that we would later add SSL acceleration hardware. But when the time came to do the detailed engineering work, we found that the general-purpose CPUs had improved enough that we could hit our specs without special hardware.  A nice surprise!
 
If you aren’t getting a lot out of hardware, and over a pretty long time, you’d really prefer to avoid it… because it does tend to lock you in.  It’s been striking to compare our software-based encryption to one competitor’s hardware-based encryption. At the point where they introduced it, their hardware was clearly ahead (indeed, at the point it was introduced we had no comparable encryption capabilities at all). But that was 5 or 6 years ago.  Now their approach is actually a competitive liability in two ways: first, the customer has to pay for the encryption hardware whether they use it or not; and second, the hardware has limited them to a 128-bit-key implementation where we have been able to support more-secure 192-bit and 256-bit variants.  As a result, Riverbed appliances can be used for information classified by the US government as “Top Secret” but that competitor’s equipment can’t – which is probably not what they anticipated when they were choosing to do encryption in hardware.
 
So where have we found a hardware investment worthwhile?  Not surprisingly, the most conspicuous performance-related examples are found on the top-of-the-line 7050 models.  In these appliances, the entire data store is built from solid-state drives (SSDs) instead of conventional magnetic disks.  Using SSD gives a major boost in performance compared to conventional HDD. That’s a good kind of “custom” hardware to build, because actually the SSDs are still commodity components.  You might think that we just dropped in the SSDs and got a performance boost, but that’s not quite right.  Even with the SSDs, there’s still “software magic” going on – we revised RiOS to make better use of the SSD-based data store.  So we’re still pretty far away from just throwing hardware at the performance problem.
 
The repeated lesson in computer systems has been that staying on the Moore’s Law escalator lets you beat custom hardware except when the custom hardware gains are huge.  At this point, we see a continuing need for flexibility and only modest gains possible from custom hardware for the most economically important workloads. So I don’t lose too much sleep over claims that some other company might be doing WAN optimization in hardware.

Posted in Uncategorized | Leave a Comment »

Remote access at warp speed

Posted by riverbedtest on May 19, 2011

Index Yesterday, InfoWorld published its review of Steelhead Mobile 3.1. It received a rave review and a score of nine out of ten.  Reviewer Keith Schultz had very complimentary things to say.  Keith notes that Steelhead Mobile takes advantage of best-in-class acceleration, the Mac version works like a champ, and best of all Steelhead Mobile just works. 

Here are some snippets from the review: Remote access at warp speed - Networking - InfoWorld

  • Instead of having to deploy a Steelhead appliance for even the smallest branch office, IT can simply push the Steelhead Mobile client to users' desktops and laptops, and they can take advantage of Riverbed's best-of-class acceleration and optimization wherever they might be — or roam.
  • Also new in Steelhead Mobile 3.1 is native support for Citrix XenApp and XenDesktop traffic. For both Citrix terminal services and virtual desktop infrastructure (VDI) farms, Mobile now helps improve client response time over the WAN and reduce overall bandwidth needs. Mobile also automatically provides QoS for all Citrix traffic, resulting in more consistent and reduced response times. Configuration is limited to choosing the port for your XenApp traffic; there isn't anything else to worry about.
  • Steelhead Mobile 3.1 now runs on Mac OS X, and it works like a champ.
  • Performance keeps pace with its hardware-based cousin and the feature set is maturing nicely, with better SSL support, Citrix optimizations, and support for Windows SMB signing and encrypted MAPI. The Mac OS X client is the icing on the cake — and best of all, it just works.
  • The Steelhead Mobile Controller gives admins an easy way to view the effectiveness of their optimization policy and the overall capacity increase on the WAN.

 

It's great to see Steelhead Mobile getting the attention it so richly deserves.  If you're wondering how to give your road warriors the same kind of WAN Optimization and performance on the road that they get when they are in the office, the answer is Steelhead Mobile!

Posted in Uncategorized | Leave a Comment »

The forecast was cloudy at Interop last week

Posted by bobegilbert on May 18, 2011

Las Vegas was a buzz last week as networking vendors showcased their latest and greatest product offerings.  Taking a stroll around the IT Expo show floor, it was obvious that cloud computing was the hot topic for vendors ranging from Intel to F5.

1

 

3
 
  Cloud_intel
 
  Cloud_f5

Cloud_cloudcommand

Cloud_force10 

Vendors were demonstrating a variety of cloud technologies that essentially help to overcome many of the obstacles that companies are facing when leveraging the cloud for infrastrusture, platforms, and software as a service.  Much of the focus of the show was around security, management, and connectivity.

Riverbed on the other hand focused on another key cloud consideration and that is performance.  As I blogged about last week, performance is a major consideration for organizations looking to take advantage of the benefits that the cloud has to offer.  Riverbed hammered home this point with a daily booth demonstration of its Cloud Steelhead product.  Cloud Steelhead essentially provides LAN-like performance when accessing servers and applications hosted in the cloud.

To further demonstrate the cloud performance point, Riverbed held a contest where we asked folks to guess how long it would take to transfer (unoptimized and optimized) a 50MB file from the cloud to the Interop show floor. 

I believe our very own Sr. Director of Marketing Miles Kelly summed it up nicely when he basically said that Interop is demonstrating that the cloud is moving from hype to reality.

 

Posted in Private Cloud, Public Cloud | Tagged: , | Leave a Comment »

March Performance Hall of Fame Winner

Posted by riverbedtest on May 17, 2011

On February 10th, a user called mikez posted an entry to the Performance Hall of Fame with the great title: "Riverbed made my LAN the BOTTLENECK!"

We love posts like that, and we're very pleased to say that a lot of other people loved it too.  It received the most kudos of any posting since it was posted, by the end of February, and so it is the March winner of the Performance Hall of Fame. 

How could you not love an entry that starts with:

"Why is our DR re-replication taking so long?" my supervisor asks.  "Is Riverbed working properly?"

After a little research, my answer was: "Yes…too well, in fact – our 100Mbps LAN is actually the bottleneck.  We need to upgrade to gig switches so that our LAN can keep up with our WAN!!"

How often do you hear a story like that? 

So congratulations and thank you to mikez for his fun and informative entry!

We'll have the April winner in a few days.  (Sorry for the delay!)  And please don't forget to enter our NEW "Riverbed Storytellers Contest."  Tell a cool story about Riverbed gear in your environment and win a super-cool iPod Nano Wrist Watch!

Posted in Uncategorized | Leave a Comment »