Hi Dexter, We will gladly collaborate on any/all of these issues if we decide to tackle them rather than waiting. AS is a small company and I'm the only one doing OrientDB development at the moment.
>From Luca's answer I see that Orient technologies have already selected full-text search as a Enterprise feature. I'm not sure how productive it would be to have another "community version" of similar features available. I understand that OT needs Enterprise-only features but I was hoping those would be focused on high-availability, management and operations rather than on "core features". I will be in touch once I know what we will do next. Best regards, -Stefán On Thursday, 20 March 2014 16:09:02 UTC, Dexter Pratt wrote: > > Hi Stefan, > > Thanks for sharing this concise summary of your findings. > > I'll chime in on one of the issues to say that our application needs > efficient text search combined with other constraints. > > I've also thought of implementing a solution later this year using one of > the existing text search options, so would be happy to hear more about your > plans. > (Our project is an open-source academic effort and I'd be glad to > collaborate on a reusable extension to share with the community instead of > building a one-off) > > - Dexter > > Dexter Pratt > Director, NDEx project > Ideker Lab UCSD / Cytoscape Consortium > [email protected] <javascript:> - [email protected] <javascript:> > www.ndexbio.org > > On Thursday, March 20, 2014 at 2:24 AM, > [email protected]<javascript:>wrote: > > Hi, > > After using OrientDB for some time now I can say that there are many > things I love about it and that I look forward to using it for quite some > time. > > There are also things that can/need to be improved that I would like to > mention but thankfully most of them are already scheduled for version 2, > due this summer. > > Please know that my only intention here is to provide constructive > feedback from someone using the product and that I'm extremely grateful for > the work that you have all done and decided to share with the community. > > Notable findings: > > - Data storage is inefficient > Highly redundant field and class names are stored for every record > (vertex/edge) and numerical data is not stored in the most compact form. > I have estimated that our data could be reduced 40-45% and that > matters when dealing with billions of records. > Work on this is planned for version 2 > > - Less obtrusive way to add edges is needed > Dealing with a constant stream of data that needs to be processed in > batches means that there is not room for countless retries to commit > changes based on optimistic locking. > I have explained this in a different thread but what we need (and > perhaps a few others) is a way to add edges without having to worry about > the status of the target vertices. > > - Sharding is not available > This is the holy grail for graph scalability :) and is planned for 2.0. > I hope the team will be implement this in an effective way as sharding > graphs is particularly tricky. > > - Better way to deal with time-series data > This my be just a lack-of-experience problem but I have found it > really hard to accommodate time-series data in a graph and a case can even > be made that such data belongs elsewhere. > I hope to provide more meaningful feedback on this once I have studied > it more but if anyone has tackled this in a good, straight-forward, manner > then please let me know. > > - Full-text search > We will start implementing a custom solution using Lucene, Solar or > Elastic Search soon and perhaps we will be able to contribute something > useful. > is there anyone here that has done successfully that is willing to > share their experience? > > - Native GeoSpatial support > GeoSpatial Analysis is in many cases a perfect fit for a graph and > having powerful, native, GeoSpatial capabilities would be welcomed. > > > I do realize that OrientDB can not be all things to all people and that > our needs and our opinions only represent one user, one use-case. > > It's clear now that we will not be able to put our Historical Event Store, > based on OrientDB, into production for our most demanding customer until > some of these things are in place but we will continue to develop it, > anxiously waiting for the next major release. > > Most of all I would like to thank the OrientDB for a fantastic job and say > that if Activity Stream was a grown-up company that I would not hesitate to > sponsor your for continued great work. > > Best regards, > -Stefán Baxter > > -- > > --- > You received this message because you are subscribed to the Google Groups > "OrientDB" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected] <javascript:>. > For more options, visit https://groups.google.com/d/optout. > > > -- --- You received this message because you are subscribed to the Google Groups "OrientDB" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
