I am looking into issue to improve the startup performance of a new instance of a Azure WebApp in a load balanced scenario.
Our website just takes too long to create the lucene indexes on the file system in a load balanced system because of large database size.
We implemented already the suggested optimizations like having the indexes saved to the local machine (sync).
Also I found this project: https://github.com/ErikHen/UmbracoAzureSearch
But in my opinion this will not fix the startup performance as umbraco will still need to create the lucene indexes for internal use in the CMS and to create images.
In short, I am trying to find the best solution that has those search indexes in a central place so new instances do not need to create the 4 indexes from scratch. As we are running on Azure and load balancing, Azure Search seems like the best solution as it also scales.
Do any of you guys have (some) experience with this? Or maybe there is a better solution and I have missed a (payed) package that is already doing this?
Can I integrate AzureSearch into Umbraco?
Hi,
I am looking into issue to improve the startup performance of a new instance of a Azure WebApp in a load balanced scenario.
Our website just takes too long to create the lucene indexes on the file system in a load balanced system because of large database size. We implemented already the suggested optimizations like having the indexes saved to the local machine (sync).
This webpage was a very interesting read as it also involved the guy behind examine : https://github.com/Shazwazza/Examine/issues/14
Also I found this project: https://github.com/ErikHen/UmbracoAzureSearch But in my opinion this will not fix the startup performance as umbraco will still need to create the lucene indexes for internal use in the CMS and to create images.
In short, I am trying to find the best solution that has those search indexes in a central place so new instances do not need to create the 4 indexes from scratch. As we are running on Azure and load balancing, Azure Search seems like the best solution as it also scales.
Do any of you guys have (some) experience with this? Or maybe there is a better solution and I have missed a (payed) package that is already doing this?
Thanks in advance,
David
is working on a reply...