Comments on: A dummies guide to the SaaS business model

By: Ob

Ob — Fri, 15 Jan 2010 01:16:37 +0000

Thanks for the simplified explanation of SaaS, it was very clear and made understanding of the subject very easy.

By: Duncan Stockdill

Duncan Stockdill — Mon, 23 Mar 2009 22:00:57 +0000

@Dennis, thanks for clarifying that you didn’t mean data loss.

Regarding scaling resilience, Twitter is a messaging system so has very different scaling challenges compared with a CRM app. I would guess you’ve seen:
http://www.readwriteweb.com/archives/making_twitter_scale.php
With a tenanted CRM system it’s relatively easy to run servers independently of each other assuming you architect the application appropriately. Add another server and you increase capacity. Amazon supports this nicely.

Paul Murphy is entitled to his view about where his data is safest. The cloud is a rather loose term and Paul’s vague concerns would appear to apply to any service where his data resides on the internet. Is this your view also?

In my previous role I was responsible for architecture of client & investment management systems in some of the biggest UK financial companies. I’m comfortable Amazon lets me create a deployment architecture that rivals the resilience of the IT systems in most of those companies. In fact I think most startups would have more success creating a robust solution on Amazon than with Rackspace say. Amazon gives you more useful tools out of the box.

Would be good to talk some time.

By: Dennis Howlett

Dennis Howlett — Mon, 23 Mar 2009 12:06:12 +0000

…and just to add a little spice into the mix, Paul Murphy chimed in on ZDN with this: “Basically, my bottom line on cloud risk is that I’ll put serious user data on the cloud right about the same day someone proves Dante’s hell a physical reality, shows that it’s frozen over, and proves that it’s run entirely by squads of flying pigs.”

http://blogs.zdnet.com/Murphy/?p=1564

We’re quality bloggers on ZDNet you know (heh)

By: Dennis Howlett

Dennis Howlett — Mon, 23 Mar 2009 12:03:48 +0000

I wasn’t talking about data loss but as it happens there have been reported EC2 losses: http://www.datacenterknowledge.com/archives/2007/10/02/amazon-ec2-outage-wipes-out-data/

I wouldn’t expect S3 losses.

I was talking about scaling resilience and the cost equation when you suddenly need to run things hard. That was one of Twitter’s problems and I know a good amount about that scenario for various reasons.

We’ve done some scale tests on ESME and while it sounds impressive to say we can run 50K users on a single souped up Linux box, of itself that’s meaningless. It’s when you have capacity problems that thing get hairy.

Then there’s the data control issue.

Amazon is fine for relatively small setups but I’ve yet to see solid evidence of a situation where I’d want to run a good sized saas solution on it.

That could easily change over time as costs fall and resilience improves.

But…that is only one part of the equation.

By: Duncan Stockdill

Duncan Stockdill — Fri, 20 Mar 2009 10:29:54 +0000

@dennis a bit bold making claims like that about Amazon without any justifying the claim isn’t it? Perhaps you would like to elaborate on these “data complications” and “spikes causing data loss” issues? As it happens, Amazon hosting is one of the most resilient solutions you can get, with data replicated to multiple physical locations in real time. As you can see here, Amazon has never lost data from their storage service: http://developer.amazonwebservices.com/connect/message.jspa?messageID=84153. It is possible for individual servers to go down on occasion as with any hosting solution, but our architecture is designed for failure so that when a server does go down other servers take over and there is no possibility of data loss.

By: Jan

Jan — Thu, 19 Mar 2009 19:13:48 +0000

Although Duane oversimplified the whole thing (i assume on purpose) the message gets through quite clearly. We have the same model at Paymo.biz and i personally am very confident in SaaS, however the path to success won’t be that easy. I see two major issues that look like drawbacks to small business desktop software users:
– cost of running SaaS over time. Many small business owners don’t realize that SaaS is actually cheaper than owning a piece of software (there are many reasons)
– privacy. Again, many customers see the privacy risks of using SaaS. I think companies need to be very careful with this aspect because a couple of negative examples could turn off a lot of “traditional” software customers. Although i’m a firm believer than a team of security experts/admins that work for a SaaS company usually do a much better job than in-house staff, i’ve ran into this issue with several customers in the past.

By: Ian Sweeney

Ian Sweeney — Thu, 19 Mar 2009 17:46:31 +0000

Duane,

Nice outline of the model. We hear a lot about the reasons it makes sense for the business (KashFLow in this example) but what is the small biz customers take on Saas? Do they give a crap? are most small biz saas users early adopters? how do they differ from the big biz Saas user?

love to hear your thoughts
Ian
billFLO.com

By: Lisa Kendrick

Lisa Kendrick — Thu, 19 Mar 2009 14:57:39 +0000

That’s the theory in a nutshell – and we’re all here to prove whether it works or not!

By: Dennis Howlett

Dennis Howlett — Thu, 19 Mar 2009 14:40:12 +0000

I should add – there is a reason why WinWeb spent £7 million

By: Dennis Howlett

Dennis Howlett — Thu, 19 Mar 2009 14:37:40 +0000

@duncan – SFdC runs its own data centers. It’s got 3 of them. And yes – I know exactly why it costs them a lot – but it isn’t remotely the number you’re quoting or as simple as how you’ve got it presented. As for Amazon – be careful…low rate throughput is fine but there are data complications as you scale plus spiking remains an issue that can lose you data. The economics are not so clear cut as you might imagine.

So while it is very easy to get started, anything running at scale is not the same, either in cost or engineering.

By: Matt Chatterley

Matt Chatterley — Thu, 19 Mar 2009 13:10:50 +0000

Good example – and its nice to see a simple breakdown of SaaS in these terms – because it’s rapidly becoming one of the most misinterpreted terms in the business arena!

By: Duane Jackson

Duane Jackson — Thu, 19 Mar 2009 12:52:30 +0000

To clarify, the £1,600 figure was plucked from thin air as it’s easy for the example of 100 customers.

By: Duncan Stockdill

Duncan Stockdill — Thu, 19 Mar 2009 12:51:13 +0000

Times are changing Dennis. When Salesforce.com started, the infrastructure would have cost way more than Duane’s £1600 per month, particularly given the scale they targeted. These days £1600 gets you 16 servers on the Amazon cloud and easy, almost infinite scalability at low cost. No doubt Salesforce.com will still be paying the price for legacy infrastructure and associated architecture decisions. Anyone got a breakdown of why each customer costs Salesforce £2000?

By: Dennis Howlett

Dennis Howlett — Thu, 19 Mar 2009 11:45:20 +0000

Hate to dampen your thunder – it’s a good post but there’s a lot you’re missing out. Oracle posted their results yesterday. 46% operating margin headed to 50%. Free cashflow in 2008: $8 billion. They’re a ‘traditional’ vendor. Highest margin position ever. Salesforce.com: saas: $1 billion in revenue, profit barely tipping 0.5%. NetSuite: saas: scraped its 1st profit in 10 years.

It’s not about whether you’re saas or on-premise (you get to profit with on-premise far more quickly when measured against capital inputs), it’s about the business model you choose to employ because that dictates scaling ability. And in saas – scale matters.