Thanks a lot for pointing me in the right direction. Seems like it is a bit dangerous to use blocking calls. i will play around and will figure things.
Thanks a lot again for your help guys. Cheers, Deepu. On Thu, May 27, 2010 at 12:05 AM, John Plevyak <[email protected]> wrote: > Just an FYI: you don't have to involve HTTP at all. > > You can use INKNetConnect, INKConnReadVIOGet, INKVConnReed > and the corresponding write functions to talk directly to the > SQL server. > > You can then use a general SQL parser and generator (one not > tied to I/O). You can also use the API to read and write to > the cache INKCacheRead INKCacheWrite. > > If you are building an HTTP->SQL gateway, you can also use the > API functions for parsing and handling HTTP headers w/o using > the HTTP engine or a transform plugin. > john > > On 5/26/2010 11:09 AM, Steve Jiang wrote: > > Hi Sandeep, > > > > A transformation plugin can accomplish what you want, but you will have > > to setup a "dummy" HTTP origin server to provide a fake response so that > > ATS has something to transform. > > > > If you want ATS to handle the request itself without connecting to an > > origin HTTP server, you can use INKHttpTxnIntercept[1] to intercept > > requests that you detect can be served from ATS. If you choose to go > > this route, you'll have to build the response HTTP headers as well as > > the body to the client VConnection. > > > > As far as connecting to the database, keep in mind that if a blocking > > library is used database requests will block other connections on the > > same thread. Using INKNetConnect with writing the protocol directly > > will allow you to talk to the DB asynchronously and continue handling > > other requests. > > > > --Steve > > > > [1] > > > http://trafficserver.apache.org/docs/v2/sdk/InterceptingHTTPTransactionFuncs.html > > > > > > Sandeep Kalidindi at PaGaLGuY.com wrote, on 5/25/10 10:39 PM: > >> Hi all, > >> i am about to write a couple of transformation plugins to > >> ATS which will fetch data from the database so that all the read > >> requests will be taken care of by the ATS itself while write requests > >> are passed on to the backend servers. > >> was checking the architecture and it looks fine by me. wanted to ask > >> the community if there is any major why we should not connect to the > >> database from ATS before i spend considerable time on it. > >> > >> Cheers, > >> Deepu. > >
