http://ayende.comAyende @ RahienCopyright (C) Ayende Rahien 2004 - 2021 (c) 202660@Henning No I dont think its rediculous. The story reads that way. Look at it this way, a client came to you to replace an existing product, you say you have a team with ORM experience, so you understand the problems and (presumably) know how to ease the friction with its use. However, you chose to go completly in the other direction, and, by using a product that you have no experience with put the entire project at risk. Great that you had a plan B, and even better that never had to be activated. You lucked out that RavenDB happens to be a great product with an active support group. You lucked out that the learning curve was not as steep as it could have been. You lucked out that the problems you faced were only minor and easy to overcome. I dont beleive that saying you had tinkered with RavenDB prior to implementing it in a commercial project counts as 'experience'. If I were the client, and when you suggested that you were going to use an untried and untested and unknown product for the storage, I would have run a mile, especially if the timeframe was tight. So yes, blind dumb luck. If you were trying to prove the technology, and the project was not highly critical, perhaps I'm being overly harsh. However these case studys are being used as tools to sell RavenDB to potential clients. I personally think RavenDB is a great product and I am in the process right now of trying to prepare a case for using RavenDB with our next product for the business I work for. I have been looking forward to reading some real world experiences to assist with this. Your story, while ultimately sucessful, does not suggest that RavenDB was the perfect fit, and would probably have been able to be sucessfuly delivered in the same timeframe using an ORM and RDBMS with a more 'traditional' approach. I want to read / hear stories where RavenDB comes to the rescue and solves some tricky and critical problems that could not be overcome using a traditional solution. http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment8http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment8Mon, 04 Apr 2011 01:19:42 GMT@Frans, @Francois, When looking over this again, I see that the index thing might seem like it was a big problem; it wasn't. It was the biggest one we had though, and that's why it made it into this interview / case study. A couple of us had tested RavenDB and had a fair understanding of how indexes work before we decided to go for RavenDB as our persistence platform, but once you grok it, it's quite easy to explain to others. It's just different. I guess you could say we traded a set of, known, old and tedious problems for a new and lesser known one, but at this point we had already decided that a NoSQL solution would be the best fit for us, so using an ORM was already way down on the list. Any NoSQL solution would be new territory for us and bring along its own set of problems. Also, this project was not just about reaching a deadline, it was about laying the foundation for a number of other projects to come, so our perspective was a bit more long term than what comes across in the interview. The decision to go for RavenDB was not an easy one, and a lot of time was spent discussing, reviewing, testing, etc. before the decision was made. To include all the reasoning behind this decision was obviously out of scope for this interview. @Me, to suggest that this is a story of dumb luck is just ridiculous. http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment7http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment7Fri, 01 Apr 2011 08:38:53 GMTI have to agree with Frans, this is not a story of success - its a story of blind dumb luck! http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment6http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment6Fri, 01 Apr 2011 02:32:28 GMTThey had a chance to try completely different technology and they succeeded. Not because they have made a series of well thought of decisions but because the technology worked in their case so the plan B could be ditched. That's nice, shows that Raven can be used in some real world situations, but it also suggests their requirements could be satisfied with a RDBMS as well. I'd like to hear more stories, especially where the technology meets some limits - of database size, of performance, of manageability etc and would also to see how Raven behaves with millions of records, thousands of users and hundreds transaction per second. http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment5http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment5Thu, 31 Mar 2011 13:35:36 GMTI think that is highly dependent on what Plan B was. Plus he said he and some of his team members already gave it a spin. I have to say, as someone who also has several years experience with ORMs, getting up and running with raven (even before dynamic indexing) is pretty straight forward and easy. http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment4http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment4Wed, 30 Mar 2011 21:15:08 GMTI'll admit i kind of saw the same thing as Frans did reading this. "We have super ultra tight deadlines, and ORMs quirks slow us down....so instead lets start exploring unknown technology!!" I consider myself pretty adventurous when it comes to technology, but when I have a super deadline, I tend to become pretty conservative... Seems like they like living on the edge :) http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment3http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment3Wed, 30 Mar 2011 17:04:32 GMT@Frans, I didn't read that into what they were saying. I read that for a changing Data Model, a traditional ORM was a pain plus he said that the developers had pains with ORMs in the past. Also, they probably didn't know about the problems they'd have with Indexes until they started the project, whereas they knew about the ORM issues that they had experienced in the past. http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment2http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment2Wed, 30 Mar 2011 15:53:10 GMTWhat's odd is that his team members are familiar with O/R mappers (so can get started right away) but they decide to go into unfamiliar territory and have a hard time learning the cornerstone of proper querying of that new technology: indexes. All this because the O/R mapping setup wasn't suitable to meet a deadline. It might be me, but that simply doesn't add up. As in: he just wanted to use NoSQL and found the reason to use it afterwards. That's fine, but at least admit it. http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment1http://ayende.com/4795/case-study-the-first-ravendb-s-production-deployment-story#comment1Wed, 30 Mar 2011 15:01:36 GMT