publishing heavy data on ArcGIS server

[AbhishekSharma1]

I am trying to publish land use data (size around 1.2 GB with 4.7 million records) on our ArcGIS server, I can divide this data in 6 different parts, still max size of data is around 550 MB with 2.5 million records.

What would be best approach for publishing this data on server, I will be using this data on open data site.

I am planning to publish this data on AGOL if my server does not support this data.

current configuration of my server is Windows server 2008 R2, Processor Intel Xeon CPU E5-2670 0 @ 2.6 GHz 2.59 GHz (2 processors),32 GB RAM with 8 cores and currently I have around 50 existing services on this server (including map, feature, GP and geocode) 

Thanks

for

[ChrisSmith7]

What about storing your data and geographies on SQL Server?

[AbhishekSharma1]

Data is stored in enterprise geodatabase (SQL server 2008 R2) 

its polygon data.

[ChrisSmith7]

You've got the database registered, so when you analyze prior to publishing, there's no warnings about having to push the data to the server?

[AbhishekSharma1]

yes, data is registered,no critical warning, no data copying to server.

my main concern is about the performance of service once its published (If it is!!) and how this will handle multiple queries 

[ChrisSmith7 · Accepted solution]

You might want to take a look at High performance web apps with big datasets as feature layers | ArcGIS Blog 

There's some info on generalizing your dataset to reduce size as well as info on how to better consume the big data within an application.

That being said, since you're not pushing the data to the server, I'd give it a go!

[JonathanQuinn]

I agree with Chris, give it a try and see how it goes.  Symbology, density of features, the number of layers, etc all have a factor in the performance and resource footprint of a service.

Something you may want to consider is caching the data so pre-created tiles are retrieved on disk instead of dynamically generating the image output.  You can still query from the service as you normally would, even if it's cached.

[AbhishekSharma1]

Thank you very much for the URL, it was indeed very helpful.