While other people are going to the free meals and executive talks with the big companies, I was spending my time looking at the companies we don't know. Looking for that one company who I thought could make a difference and, I think I might have found it. In the little row called Startup where other members of the press came by and gave companies five minutes, paid no attention and left. I spent 45 minutes asking every question. I am very sure I overwhelmed the staff of the 6 foot booth but sometimes you have to ask the hard questions.
So what is Fastsoft? That is the same thing I was saying when I visited the small booth on startup row, looked like another wana be player in the WAN Acceleration movement. But what I found was that they have a place and they will not compete with Cisco, Riverbed or Juniper in the space. Fastsoft only needs one box at the data center; no end box to buy and no agents to load on the far side. These boxes start at around 10k and go to 80k. They will not reduce your bandwidth like other players; you will not get an ROI back that way. What they do is use a self designed FastTCP protocol to detect congestion on the line. Then it makes adjustments so the line stays smooth without the bursts.
They can take a 25GB Blue Ray file on a 1Gbps pipe from New York to Los Angeles in 22 minutes where it was taking 45 minutes. Now let's look at New York to London with a 700mb CD File that took 1 minute instead of 14 minutes. Another good test was Los Angeles to London over a 100mbps pipe sending a DVD Disc file of 4.75GB. It took 9 minutes instead of 128 minutes. The results are good so far so what could be next and where could you see this helping you?
Netflix would be a great company to use this, Apple and ITunes would be another great bet, any company wanting to speed up the delivery of the content in audio or video formats. We will have to ask Fastsoft to send us a box for testing, but all looks bright for this young startup.
Advertisement: |
Larry Chaffin Ph.D is the CEO/Chairman and founder of Pluto Networks, a consulting and VAR partner specializing in WDS, VoIP, WLAN, Telepresence and Security. Pluto Networks is a leader in WDS-Application Acceleration, Full Disk Encryption, End Point Security and Telepresence. While specializing in the needs of large and enterprise companies, Pluto Networks has been concentrating on the SMB customers to provide them with the same great service as larger companies. Pluto Networks holds SMB specializations from our partners to service all their needs. Pluto Networks has become a leader in SMB VOIP using Cisco and Linksys to service customers.
Managing Cisco Secure Networks, Skype Me, Practical VOIP Security, Configuring Check Point NGX VPN-1/Firewall-1, Configuring Juniper Networks NetScreen & SSG Firewalls, Essential Computer Security: Everyone's Guide to Email, Internet, and Wireless Security, How to Cheat at Microsoft Vista Administration, Microsoft Vista for IT Security Professionals, Asterisk Hacking, 2008 VoIP and Video Conferencing, Infosecurity 2008 Threat Analysis and author of Building a VOIP Network with Nortel's MS5100, along with co-authoring/ghost writing eleven other technology books for VIOP, WLAN, security and optical technologies. Larry is currently working on a follow up to Building a VoIP network with Nortel's MCS 5100 Book as well as new books on Cisco Telepresence Networks, Practical VoIP case studies and WAN Acceleration with Riverbed.
Larry has more than 29 vendor certifications and has been working on many others. Larry has been a principal architect around the world in 22 countries for many Fortune 100 companies designing VoIP, security, wireless and optical networks. He has expanded over time also to include application acceleration or WDS. Larry is working with major vendors now on updating current certification tests to make them real world focused.
The Leader in Network Knowledge
Any software needed on the remote hosts?
Is there any software needed on remote hosts? Like a new FastTCP stack for each Windows, Linux, and MAC client?
Response
Michael,
In the wikipedia link http://en.wikipedia.org/wiki/FAST_TCP
It explains that Fast TCP is compatible with existing TCP and only requires tweaking the TCP settings to the computer which is sending data.
-CCNP40
Aspera Benchmarks in Comparison
The three cases cited in the article work out to be
162 Mbps for a cross-US transfer on a 1 Gbps link
97 Mbps on a NY to London transfer on a link of ? Mbps
75 Mbps on a LA to London transfer on a 100 Mbps link
Please note that Aspera software is already deployed by both iTunes and Netflix for receiving video content, and achieves much higher transfer speeds for the same network conditions and bandwidth capacities (without employing compression or caching).
For example, AboveNet has demonstrated California to London transfers on a 1 Gbps link at 795 Mbps using Aspera software, and many Hollywood studios and their partners regularly transfer to iTunes at speeds exceeding 100 Mbps from many different points in the US. Aspera's fasp technology is the standard high-speed transfer protocol used by iTunes to receive the large majority of their video content on a daily basis.
The fasp protocol is targeted at bulk data transfer applications ("file transfer"), and unlike Fast TCP, is designed to tolerate extreme packet loss and delay and to achieve the fastest possible transfers on commodity hardware, even on global WANs. Software is required on both endpoints as the protocol is application-layer, but this has not been a barrier to deployment when embedded in the software application responsible for transfer.
For more information, please have a look at our web site: www.asperasoft.com.
Michelle Munson
President and Co-founder
Aspera, Inc.
This thing is getting big
Larry,
A bake-off between FastTCP and Fasp would be more interesting than the one proposed between WAAS and Steelhead for these reasons:
1. Solutions calling for boxes at both ends of a link are something in the WAN space for over ten years beginning with vendors like Packeteer, something boring, uninspiring and technologies well digested by your smart blog readers;
2. Single-ended solution and software embedding are new to us, and in my limited knowledge of the market, something revolutionary;
3. Both companies target the same set of customers;
4. This will let you clear your name in front of those accusations that your blog lip-serves only the big house and Riverbed;
5. Despite of the fact that the market is aready very crowded, there is still a large portion out there with environments where two-box solution simply unfit.
Fastsoft be the next big thing?
This sounds interesting, but first some qualifying questions to be answered:
1. Was the data quoted from some Powerpoint slides by Fastsoft, or it was observed on-site at Interop in real-time? If latter, how?
2. Quoted test results seem to derive out of three data centers in New York, LA, and London, were there boxes at all test locations at the time of testing? If yes, we are talking about symmetric acceleration here, NOT what it sounds like an asymmetric solution.
3. If it's a symmetric solution, I have to disagree with the notion that Fastsoft is not in competition with the other players. Do we have an internal test bake-off result comparing with Riverbed, Citrix WANScaler, SilverPeak etc. in a similar environment?
4. If it is an asymmetric solution, the results are very impressive indeed, but were the tests run on end-to-end International leased circuits, or on public Internet?
5. Were artificial latency and packet drop inserted?
In order to qualify for "The next big thing", the newly invented technology involved must address a market with width and depth. I believe the number of targeted customers, who can afford a London - NY fat pipe, is very limited, not discounting those who are already the customers of the existing players whose TCP acceleration products can do similar jobs.
Are we seeing another player jumping into the crowded enterprise WAN-OP market? If yes, it will not be a big thing in my judgment.
FastSoft asymmetric solution: one-sided TCP acceleration
As a VP at FastSoft, I can answer these questions. The data was taken from our real-world measurements between data centers and users. The list of cities, file sizes and speeds in our data table was selected to be illustrative of the kinds of benefits that FastSoft users experience; all quoted results were run on the public Internet, and there was no artificial latency or packet drop inserted.
We have an asymmetric solution that is different from the two-ended solutions that require hardware or software at both ends. We do not rely on caching or compression, as most two-sided solutions do.
As for your point on WAN-OP, you are 100% correct - FastSoft overlaps much more with the rapidly expanding Content Distribution Network (CDN) market than it does with WAN-OP devices. In fact, FastSoft actually complements CDNs by helping solve the intractable last-mile problem!
FastSoft asymmetric solution: one-sided TCP acceleration
Thanks Product Management for the answers. The solution sounds exciting!
It would be very nice for the blog readers here to get a first hand experience of the claims. Unfortunately Reuters Australia (your customer declared on Fastsoft website to have implemented video acceleration solution of Fastsoft) doesn't seem to have a public video site I can get a taste of the accelerated streams and compare that to say, Reuters Canada's streams. I have a 6Mbps (download)ADSL. When I watched Reuters U.S. (assumed without Fastsoft)video streams I got average 550kbps buffering speed. If the technology works well I anticipate to get 1.5X performance improvement from Reuters Australia on the conservative side, and according to your whitepaper.
Do you have another video site that we can experience this exciting technology?
wopro