Tuesday, July 18, 2006

SaaS is letting me down

I'm getting pretty pissed off with my SaaS suppliers at the moment. In the main, that is GMail, Salesforce.com and Blogger. My main issues are twofold:

1. Slow performance
Dave Stephens recently gave his commentary on a recent Aberdeen report on Procurement SaaS. A key comment caught my eye because it resonated with my own experience:
“The only real negative I could find was in one statistic: 88% of respondents said that in house software system response time & performance was equal or better than SaaS. This reinforces my SaaS = renting-an-apartment analogy. At peak times some customers’ showers are running out of hot water! But it’s a solvable problem - the SaaS providers need to design their on-demand systems for peak load, not average load. And hopefully they will start doing that.”
Start. Doing. That. Well, yes, could they please get off their backsides and take care of this!?

[EDIT: Phil Jones rants more about Blogger on his Platform Wars blog. In the comments to his post, John from FreshBlog mentions the Blogger Hacks Wiki as a good place to find how others are 'hacking' Blogger. Heard about Phil from Dave Winer's Scripting News.]

GMail was wonderful when I first started using it, but in the last few months it has slowed to a crawl - even on a fast internet connection I find that simply labelling a post in my inbox takes 10-30 seconds. Salesforce.com has had some very public issues with its performance, which their main competitor, NetSuite, blames on their 'big iron' architecture. But the bigger issues for them are over-reliance on an old web interface paradigm, that in these heady days of AJAXified web apps seems a very inefficient way of forcing users to work with their data. Blogger has the least problems in this regard, because of its nice good UI design and their simple technical model (i.e. publishing your blog as static web pages to maximise cacheability and performance).

It's nice to know I'm not the only person getting annoyed at this.

2. Gluggy feature sets
gluggy slow, sluggish - used to describe the operation of a computer”
Urban Dictionary
Blogger is the biggest culprit here. They've managed to add a handful of features to their blogging application in the last few years, and as a result now offer a far less satisyfing feature-set than many of their competitors. It is only thanks to the strength of the original design (pre-Google I might point out) and the great user-community, that people who want extra features (like categorisation) still hang around. The time must come soon though, when people start to move elsewhere, especially now that China, India, Pakistan and Ethiopia have taken to blocking access to blogger blogs.

Salesforce.com is bad in a different way, they have made me sluggish. Yes, AppExchange is a wonderful idea, but did we really need to get drowned with a thousand different applications? There is something to be said for having our choices restricted to a few truly great options, rather than the mess of good, bad and ugly that AppExchange really is. Once I wasted several hours looking at different applications, I realised that none of them would offer me anything I really needed or wanted on top of the basic Salesforce.com functionality (although I hear that one free one offers RSS feeds, that might be worth looking into). I could have better spent that time getting more use out of Salesforce.com's existing functionality rather than surfing the AppExchange Deadzone.

GMail has added some features, and steadfastly ignored others that users clamor for (like sorting by sender, or date, within a label or search view). The integration with Google Calendar is painful and half-baked. The fact is that my own use of it relies heavily on Greasemonkey, and so I find that using it from within IE (as I usually do outside of home) has become oddly stilted, as I reach for functions and options no longer available.

In Summary
I am a strong believer in the viability of the SaaS model. Removing maintenance headaches and installation gotchas are worth a lot, but the efficiences gained in being accessible anywhere and having economies of scale in the backend make it really appealing. I find myself agreeing with Jeff Kaplan on his identification of SaaS Myths, but in reality the applications on offer are still not fully ready for the enterprise. Many of them have not yet been tested on their performance, or security, or ability to keep up with technology changes. Market leaders like Salesforce.com, GMail and Blogger are showing signs of not handling things as well as they should, perhaps because the SaaS technical model is not yet fully mature, or simply because there is yet to be a realisation that they must handle peaks in demand, not just the average.

Technically speaking, SOA and advances in grid computing promise to deliver new levels of performance scalability, but applications must be built from the ground-up to full utilise these advantages. However, this sort of application programming can be pretty tricky (and yes, I know that threading <> grid computing, but it's got similar problems). However, there seems to be a recognition that this is an area of great opportunity, and some new vendors are planning to supply this next generation architecture.

From a security standpoint, enterprise IT departments have fundamental issues with hosting their data outside the corporate firewall. We have seen a recent rise in online storage providers, with ones like Amazon's S3 being aimed squarely at developers needing back-end data storage. Perhaps the next wave will provide data storage that is transparent to the SaaS application, but where hosting can either be outsourced or kept in-house. After all, SaaS providers usually have to promise not to access your data anyway, so why should they be bothered hosting it?

I think that the current depictions of SaaS 2.0 are missing the point if they think that better SLAs are all that is needed. Vital issues, such as performance, require both that the application be built differently from its very foundation upwards, and depend on the rollout of faster broadband access speeds across their marketplace.

For more on this subject see Dion Hinchcliffe's blog. But from my POV, the tipping point he's looking for ain't here yet.

del.icio.us tags:


  1. Anonymous6:08 am

    I think a lot of the slowness issues are like you said, due to over-ajax usage. While ajax is a cool and useful tool, if you use it too much you are constantly downloading additional data. There are reasons to leave some features ajax free...

  2. Ross,

    I think you're right, it's also a matter of how much they cram into each page.

    I've noticed that minimising those areas on the left-hand side (labels etc) improves the page performance - so I'll be doing that from now on.

  3. SaaS still has along way to go and it will all be a work in progress or perpetual beta as some would call. Your points illustrate them well so aa to be taken as learning curves but not at the risk of completely doing away with it. Maybe we just need to be patient for real SaaS champs to emerge incorporating all that is best about technology that truly address needs.