Catching up with Ariba's Craig Federighi on Open Source
Since Coupa, an Open Source eProcurement vendor, announced their initial product release, I have spent some time reaching out to a few senior folks in the Spend Management technology world to get their take on Open Source and Spend Management. This post features notes from my discussion with Craig Federighi. Craig, Ariba's CTO, is probably the most lucid and learned technologist in the sector that I know. In the next couple of weeks, I'm hoping to toss in some alternative perspectives from Ketera, Emptoris and possibly others on the subject as well.
According to Craig, Ariba is a big proponent of Open Source from a development and platform perspective. But on the application front, Open Source presents "a great unknown as to how it will pan out." Today, Ariba extensively leverages the Open Source economy by embedding different components and code within its applications. This might take the form of XML parsing libraries, Apache, Tomcat, or Lucene search. In this regard, Ariba "benefits from Open Source in the exact way an Open Source vendor does. Because developers use these components across a wide array of application verticals, there's enough developer energy around them to constantly enhance the code." The closer you can get to a lowest common denominator used by millions of developers -- such as an application server -- the greater the benefit for the Open Source community at large. This is due to the fact that there's a tremendous amount of energy for free that wants to iterate on, fix it and improve Open Source code at this level.
But Craig feels that it's unlikely that Open Source will make great inroads in the eProcurement application market further up the stack. On the business applications level, you don't get the "multiplier effect of free engineering". Sure, there might be niche areas like approval engines and scripting layers for writing extensible business rules, but in general, Craig believes that the Open Source developer effect won't materialize on the Spend Management business applications level. In addition, there's also the separate question if businesses would trust Open Source code. The fact that eProcurement is part of a financial transaction -- and touches internal transaction systems -- may scare Open Source prospects. What happens if someone slips something in the Ruby Code?
Getting beyond the security issues, there's a big question about whether Open Source business applications make sense financially. That's because when "you look at the cost of a traditional enterprise application, the license cost for installed software is typically a fraction of the installed total cost of ownership." The question becomes how much will I pay Accenture to implement it, Oracle for the database, and the internal IT team to manage it? Ultimately, the additional costs can account for 6x the license costs (making a non-Open Source component a small piece of the overall cost). Compare this with an On Demand approach from a single vendor which has the economies of scale to operate software as a service with a high quality of service without the type of internal costs that come with running software behind the firewall. You don't have to employ the DBA, set up the fault tolerant networks and the disaster recovery center -- all of this is done for you.
Craig also believes that when it comes to eProcurement, there's a whole host of hidden costs that the software license alone does not address. These include content management and supplier management among other areas. In other words, the license component of eProcurement is small relative to the total costs of running and supporting the application over the long haul -- and that's before you even get into the feature / function discussion.
Spend Matters would like to thank Craig for taking the time to share his thoughts. If you are a senior technologist or executive at a Spend Management software or services provider and would like to share your opinion on Open Source for publication, please get in touch: jbusch@spendmatters.com
- Jason Busch
According to Craig, Ariba is a big proponent of Open Source from a development and platform perspective. But on the application front, Open Source presents "a great unknown as to how it will pan out." Today, Ariba extensively leverages the Open Source economy by embedding different components and code within its applications. This might take the form of XML parsing libraries, Apache, Tomcat, or Lucene search. In this regard, Ariba "benefits from Open Source in the exact way an Open Source vendor does. Because developers use these components across a wide array of application verticals, there's enough developer energy around them to constantly enhance the code." The closer you can get to a lowest common denominator used by millions of developers -- such as an application server -- the greater the benefit for the Open Source community at large. This is due to the fact that there's a tremendous amount of energy for free that wants to iterate on, fix it and improve Open Source code at this level.
But Craig feels that it's unlikely that Open Source will make great inroads in the eProcurement application market further up the stack. On the business applications level, you don't get the "multiplier effect of free engineering". Sure, there might be niche areas like approval engines and scripting layers for writing extensible business rules, but in general, Craig believes that the Open Source developer effect won't materialize on the Spend Management business applications level. In addition, there's also the separate question if businesses would trust Open Source code. The fact that eProcurement is part of a financial transaction -- and touches internal transaction systems -- may scare Open Source prospects. What happens if someone slips something in the Ruby Code?
Getting beyond the security issues, there's a big question about whether Open Source business applications make sense financially. That's because when "you look at the cost of a traditional enterprise application, the license cost for installed software is typically a fraction of the installed total cost of ownership." The question becomes how much will I pay Accenture to implement it, Oracle for the database, and the internal IT team to manage it? Ultimately, the additional costs can account for 6x the license costs (making a non-Open Source component a small piece of the overall cost). Compare this with an On Demand approach from a single vendor which has the economies of scale to operate software as a service with a high quality of service without the type of internal costs that come with running software behind the firewall. You don't have to employ the DBA, set up the fault tolerant networks and the disaster recovery center -- all of this is done for you.
Craig also believes that when it comes to eProcurement, there's a whole host of hidden costs that the software license alone does not address. These include content management and supplier management among other areas. In other words, the license component of eProcurement is small relative to the total costs of running and supporting the application over the long haul -- and that's before you even get into the feature / function discussion.
Spend Matters would like to thank Craig for taking the time to share his thoughts. If you are a senior technologist or executive at a Spend Management software or services provider and would like to share your opinion on Open Source for publication, please get in touch: jbusch@spendmatters.com
- Jason Busch
First, I think it's great Ariba is leveraging open source componentry to benefit their customers. Lucene, as an example, is a credible open source search engine - and it's faster Ruby implementation, called Ferret, is something Coupa employs. So if you're an apps vendor and you're not using open source components you are probably spending way too much on development.
Obvioulsy my views on whether open source makes sense for enterprise applications differ strongly with Craig - and so the market will get to decide. I think it's fantasy that somehow open source has crashed through all these barriers but then will be turned away at the apps layer. But we'll see.
On TCO, I certainly also hold the view that license is a fraction of total implementation cost. I want to quickly add customers are very tired of hearing that argument though - as it is usually a vendor justification for $900k annual payment (or perhaps monthly subscription payments for $75k). So Coupa is focused on lowering up-front *and* ongoing costs. And I definitely think SaaS can help there but it's not nirvana.
Back to open source, I do have an incremental suggestion for improvement to give Craig & Ariba. An easy & beneficial step towards being more open is to give up control of the cXML protocols. My prediction if you don't is that it's likely it may not last another 5 years. And if you do make it open chances are it will last longer in the market & you will still probably have the best support for the protocol of any vendor.
Most of all & in closing, I appreciate Ariba weighing in on what my young company is up to. It's a sign we may be on to something... :)