I promised earlier to give my own thoughts on the Dmitry Goroshevsky interview and Peerio, so here goes. As a courtesy I passed the text on to Dmitry in advance of posting, and his “right of reply” comments are at the end.
Now, as it happens some water has flowed under the birdge since then, with various press releases, patent applications and comment. So I’ve got a bit of free hindsight to make me look wiser than I really am, or at least not make predictions that are immediately proven false.
I’m going to divide my thoughts up into three bits: the business strategy, the technology, and the value offering.
Business strategy
I was a bit disappointed by the lack of clarity on who the customer is, what their need is, and how the solution is going to be sold to them. Likening yourself to the Microsoft monopolies sets a dangerously high set of expectations. Is the customer really the device manufacturers? If so, what’s their unfilled need? I still think that in larger businesses the telco manager will note that this puts him out of a job, which will limit adoption. For big enterprises the “PBX” functions are also a side-show compared to the custom call distribution logic they’ve built. Avaya and friends make application platforms, not PBXs. A PBX replacement isn’t a threat, because that isn’t where the cost is to the customers. I don’t see a P2P architecture making much of a dent on that customisation and integration cost.
Not knowing who the customer really is need not be fatal; I doubt Microsoft really understood the full dynamics of Windows and Office until relatively late in the game. As a technology vendor, Popular Telephony has the luxury of allowing it’s partners to carry much of the cost and risk of market discovery, just like the PC vendors acted as channels for Windows.
Technology
Some of you will know I was a consultant at Oracle for a number of years in the Server Technolgies group. Before that I built Informix-based transaction systems. So I’ve got a professional interest in data stores.
Big iron databases exist for a reason; when you want to mix’n’match large datasets, it helps when you do it on a tightly joined set of CPUs, memory and IO buses. Highly distributed databases are OK for higher-latency applications, of where the data doesn’t need much post-processing after capture. Peerio’s persistence layer therefore only applies to a relatively narrow set of problems, such as the storage of binary objects like voicemails. If you ever need to run complex reports on your directory or call patterns, you need an old-fashioned big iron server.
Dmitry’s claim about the limits of scaleability of centralised systems clearly has some truth to it. BitTorrent exists as much for economic as technical reasons. However, I’m skeptical that this is a serious issue in a telephony application for the near future. The quantities of data aren’t huge, and within an enterprise it doesn’t really matter about balancing the costs of hosting and consuming data.
Value offering
The value offering is where things get interesting. The users don’t care what your business strategy is, and (I maintain) don’t really care what the technology is, particularly in the SME segment. So despite my doubts on both the business and technology strategy, this could turn out to be a significant hit. The value proposition is much like Skype’s — it just works, and it does it cheaply. I’m also a fan of the device-centric approach that allows experimentation in the enterprise without having to seek permission from IT department gatekeepers. That said, I got told off once because I was running Skype and it was port-scanning for supernodes; Peerio might come up against the paranoid centralised security bureaucracy of large enterprises.
As pointed out elsewhere, the distributed data system could be a headache if the network topology gets altered radically. When I worked at Sprint, the company relocated 15,000 staff to a new campus from locations all over Kansas City. I wonder how well a Peerio solution would cope with such radical upheaval.
Summary
It’s pretty brave to get Cluetrained and be interviewed by a blogger. I’m not being paid to say anything (no, really), and don’t have any advertisers who might get upset if I’m too critical. So I’m pleased they took the chance.
I think the serverless concept is a red herring; the important thing is self-configuration. If there were servers in the enterprise that could be automatically discovered and used, there’s no reason not to do so. The most important thing for Peerio to get right is the management interface; ideally, you should never need to see it.
The idea of “revolution from the edge” appeals to me still, and the idea of being able to buy devices that meet the user’s needs, yet still play nicely in the corporate sandbox, is one worth persuing. From that perspective, I’m still a fan.
If they do make ease of adoption and low management cost their focus, they’ll have a very compelling offering to medium-sized businesses. Get lost on the technology, and it’ll be a footnote in the history books on communications. As for the wider adoption of Peerio as a distributed computing platform, we’ll have to wait and see — I’ll reserve judgement on that one.
Dmitry’s response
The only thing I can say is that all the concerns you have are there of course. But the same concerns were there for mainframe and client/server architecture, so what? One concept is good for certain thing and not so good for another, for example Mainframe is far from being dead, right? Transactions are still better on mainframe. And servers will be there for dynamic web content for example (which is something I do not think serverless architecture need to replace). The real question is what will be a MAINSTREAM architecture for enterprise computing. Here, I think, if I’m taking all pros and cons (being it what you wrote and more) I still counting a clear 10:1 victory for my approach. It is cheaper, easier, better for the network and more flexible. But I guess we will see who is right about it. I can tell you only this - when you come to the Company and saying you can save 80% of the cost on 80% of all what you need - this is not “appealing”… It is HUGE. IP-PBX will be replaced, but not only IP PBX (btw watch for VON announcements from PT :) I’m serious!
Dmitry Goroshevsky
ceo
Popular Telephony
TrackBack URL for this entry:
http://www.telepocalypse.net/cgi-sys/cgiwrap/mgeddes/MT/mt-tb.cgi/413.
Listed below are links to weblogs that reference Peeriooooh:
»
Peering at Peerio from Aswath Weblog
Popular Telephony has developed a serverless computer/communication technology called Peerio. They have talked up the benefits of this technology without giving much detail, either on the specifics of the technology or on the specific advantages. The f...
[Read more]
Come-on, there's a lot people involved here and there's a lot at stake. If you want to make money under these circumstances, you have to be inclusive to at least a majority if not a super-majority. Let's be realistic about the realities of a pervasive edge network design...
Posted by: at March 5, 2005 03:19 AM