|Professional Services||Second Signal||Presentations||Andrew's Blog||Support|
So Microsoft is buying Skype and a lot of people are questioning the reasons and the price. While the price does seem high, virtually all tech world buyout prices seem ridiculous to me and frankly, Microsoft has a ton of cash laying around that they need to invest, so I'm going to table the price discussion and let the wall street types argue over that part.
From a tech perspective, if you start by agreeing that voice and video over the internet is going to be important over the next decade, you're halfway there. Microsoft has long held the belief that this is a space they need to own. They've got their "Live Communications Server" which is a pretty strong offering that nobody outside the tech world knows about. They've got their Messenger chat program that has a very small active user base compared with AIM and Skype. They're just not getting through noise and making their products lead the market.
Lets take a look at what makes Skype such a power player for end users today:
#1 - Connectivity.
Skype just works.
Skype uses a different (and proprietary) communications protocol that works extremely well through routers and firewalls from the cheapest home appliance types up to the most sophisticated corporate configurations. To users, it seems to just work. The voice, video, and chat capabilities don't require any custom configurations to talk to the rest of the world -- and in fact are fairly hard to block at the firewall even if you want to block them. There are down sides to Skype's methods of connecting and use of bandwidth, particularly from a corporate perspective, but they are invisible to the end user.
Most of the corporate world -- including IBM Lotus Sametime and Microsoft Live Communication Server -- uses SIP and/or h.323 for connections. These are very rich connection protocols, but they have major problems with firewalls. In these protocols, the connection is handled using the established connection but then data is transferred over an entirely different port which gets negotiated at the time of connections. Inexpensive firewalls are simply not able to follow this kind of connection, leaving end users trying to use the systems from home without audio or video content. Rolling out a SIP or h.323 based solution to end users at home requires significantly more support and often hardware upgrades. This is why Vonage sends out their own router as part of their service offering.
There is, of course, an excellent open source alternative to SIP and h.323 called IAX2. I'm using it for my Second Signal work. IAX2 has had a slow adoption curve and has had a hard time matching the scalability of SIP. By including the audio and video data within the protocol path itself, IAX2 has a harder time reaching the kinds of massive scale that the other protocols reach.
#2 - Clarity.
Skype "just works and sounds great". The reasons for this are complex, but I'll try to outline it below.
Skype again uses a proprietary CODEC to compress voice data. A CODEC is the definition of how you compress, store, and un-compress to playback the sound or video (or anything else). A telephone in the USA uses a CODEC called "ulaw" which is very simple and not compressed. The audio data is limited by only carrying a limited frequency range of sounds. Regular telephone data runs about 60k/second. Your cell phone typically uses GSM for the CODEC (not to be confused with the other "GSM" which deals with the connection and radio signal and competes with CDMA -- that's a different GSM). GSM also has a limited range of frequencies but compresses the data to about 13k/sec. It's a fairly simple compression that older cell phone microprocessors can easily keep up with. The down side to GSM is that you can hear how much less voice data actually gets transferred. It's why your cell phone doesn't sound as good as your land line. VoIP telephones (like Vonage, or your digital office phone) have a variety of CODECs to choose from. The CODEC used for the call is negotiated by the connection protocol when the call is made. Both ulaw and GSM are options, but there are several others. One of the most common of these is called g.729. The advantage of g.729 is that it has a much wider frequency response than even ulaw -- it picks up deeper low tones and higher high tones -- but it compresses them very effectively, using only about 9k/second. The down side to g.729 is that it is a complex compression algorithm and requires much more processing power to keep up. It sounds terrific to the users. g.729 is also a patented protocol and to use it you have to pay a small fee per connection. This has to be built into the cost of phones, pbx units, and in some cases software. That's expensive if you want to give away your chat software, which is why Skype uses their own proprietary CODEC. I've heard that it is similar to g.729.
When you pick a CODEC for sending voice traffic, compression is important. A cell phone company can handle almost 5 times as many calls over the same radio frequency using GSM than they could with ulaw. That's a huge cost savings. This holds true for your internet connection, but there's another issue. If the data is sent at 60k/second and your internet connection is doing anything else, many home internet links can't keep up. They have plenty of download bandwidth, but not much upload room. If your connection can't keep up with the needs of the CODEC, the sound will degrade terribly.
#3 -- Persistence
With Skype, users can log in using multiple computers and cell phones at the same time without interfering with each other. Chat and phone calls will simply route to all of them at once. In the case of a phone call, whichever one picks up gets the connection. When multiple users are in a chat, as long as any one of them is connected, the chat can persist. When others log back in, they catch up with the outstanding updates. This makes for a very effective "on the fly" community. It accomplishes much of what Google was trying to do with Wave, on a much simpler basis.
What this means for Microsoft
With Skype, Microsoft gets the ONLY successful VoIP alternative that users have accepted, that works through firwewalls, and that can be purchased. Nobody can "own" SIP or h.323. IAX2 solves many of the problems with those other protocols but it's open source. Open source presents many problems for Microsoft because they can't just change it the way they want without sharing their changes and they can't keep competitors from duplicating their work.
Microsoft buys a proven technology that users like, and one that gets very good press. Skype is de facto standard for cheap and easy international voice and video communication because it doesn't require any real effort or preplanning to set up. It just works. If you have a connection, you can get skype to work.
Where Microsoft will get into trouble with Skype
While Skype works amazingly well from an end user perspective, that very connectivity presents problems for Corporate firewalls. First, Skype is very hard to block. It can be done, particularly with expensive and complex firewalls, but it's not easy. Second, Skype's communication is a kind of peer-to-peer hybrid model. When you connect with Skype, you have NO IDEA what route your data will take to its destination. If you're using skype, your own system becomes part of the routing network for the global Skype data. Corporations, with their very fast (and expensive) network connections, can accidentally become "Supernodes" on the Skype network, carrying data for many other people. The protocol for this is very secret. Skype has never been willing to fully share the details. This kind of connectivity works extremely well in an open and free client for end users, but once Microsoft is behind it, the game changes. End users at home are going to resent carrying data for other people in corporations. Changing the connectivity protocols in Skype would be messing with the very thing that makes it successful. Microsoft will have a real challenge with this.
Please wait while your document is saved.