Article posted on Jan 16
Last friday I attended a CTO function put on by the UTC. The presentation was by Alex Barnett of Bungee labs who was speaking on “Trends in Software as a Service (Saas).”
The turnout was very impressive. There was probably 50 or so local CTO’s in the room. The interaction was intelligent an interesting. The presenter was entertaining and well informed on the subject matter. Honestly, the only thing I think I would have liked to have seen more of was a presentation on what Bungee is actually doing in this space. Bungee is doing some really cool stuff, and developing some great tools to allow developers to leverage the new “componentized” trend of SaaS inside their applications. Alex was so respectful of being objective and not hi-lighting his own company that I think we missed out a bit on those gems. But it was still a great event.
My personal opinion on Saas: I think we are currently6 seeing a whole new trend in Saas… whereas in the past Saas has basically meant “My application can be used online”, and has since expanded to “You can get and put some basic data through my online API to sort of integrate with me,” I think we are now seeing this being to become more granular, and companies are building “building blocks” as services to allow them to be part of a solution. This is very very cool, but I see some hurdles:
Cool side: This allows companies to build things very quickly. There’s no need to wait for hardware, or order servers. It’s all just there. Additionally, it allows for scaling up very quickly if needed. The new services offered by Amazon as an example, even allow an application to “self provision” more copies of itself. So if you get hit to hard your application can actually spawn onto more servers as needed.
Concerns: Pricing is definitely one. I think companies will need to be very careful to look at when these building blocks become more expensive than just doing it yourself. It may well be that they make sense in the early stages, or in an application that hard huge periodic spikes in usage, but it may be less cost effective in a mature application wit more consistent needs. Additionally, I think there will be a la of a year or so in the acceptance of the VC world to fund companies who have their own IP that is then reliant on the “building blocks” supplied by other companies. It can be a bit scary when a key piece of your IP is so very much out of your control. I also think that companies like Amazon are really going to how to prove that they are reliable… as much as Salesforce.com is touted as a great example of Saas, I can tell you from personal experience that the reliability of the APIs at times have been terrible.
As a note, I am also really looking forward to the event later this week on New Media and where it is headed.
Steve - thank you for the kind word re: the presentation. I really enjoyed the quality of the audience - fun to do.
Re: the Bungee Labs pitch, or lack of. Sorry I didn’t go more into that…Yeah, it’s cool stuff. I would be very happy to walk you through what we’re doing (including how we help developers programming against Salesforce services) and to get your feedback. If you want to follow up, mail me: alexb at bungeelabs.com
Re: the pricing concern and the example you gave: “but it may be less cost effective in a mature application with more consistent needs.” Right…but imagine if it *were* price effective? Would that be interesting?
Re: your IP point / VC point: You wrote VCs would be retisant “to fund companies who have their own IP that is then reliant on the “building blocks” supplied by other companies. It can be a bit scary when a key piece of your IP is so very much out of your control.”
Here’s the thing: VCs have been investing in companies that rely on third party “building block” for years, they just need to realize it. Many, if not most software startups (inc. one-person startups / entrepreneurs) build their IP over licensed “building blocks”, licenced to them by other IP based companies.
Think of the database as an example. Many software startups build apps that require a relational datastore rarely build their own db engine themselves. It’s the schema and the data sitting in db where some of their IP value will live. There is a well established reseller licencing business model based upon this fact. Service-based models exist today (and new ones will emerge) allowing and assuring a clean separation between developers’ IP and the IP provisioned via a third party service, which in turn is part of the overall service the developer has developed. I do agree with you it will take time for mental shifts to occur in this point.
On reliability - I agree, this a key.
By the way, I wrote up my presentation notes over at my blog where you’ll find pointers to some of the resources I mentioned.
Thanks,
Alex.
Alex, great to hear from you!
Real briefly in reply:
More info… I will definitely take you up on it.
Pricing: Agreed. I hope it is the direction things go.
VC: I completely agree. So many aspects of IP are reliant on outside components, from power, to network, to servers, to (as you pointed out) databases, JDKs, Sendmail/Qmail/Linux, and so on. But I see us at the point right before linux got acceptance. Back in 1999 I sold my company that used linux extensively… but when they did diligence we had to be sure to focus discussions around our SUN components, because they were more widely accepted as “ok”.
I really believe that VC’s will accept these outsources new pieces as a valuable part of a solution. I’m just saying that, like it was with linux for a while, they may prove a little bit tentative on this issue for a year or so as it gains more popular acceptance.
As to reliability, I’m glad we both agree… we should start a company that tries to provide that… oh wait, that’s exactly what you guys are doing