Hello everyone
I'll have to deploy an arcgis entreprise on a server. However ... first things first! I need a server and I do not all information I need to dimension the hardware. And I might be completely off as it is a premiere for me to deploy that kind of inrfastructure. I need advice from experienced people because I'm a bit lost with the ESRI documentation
The only constraint is a single machine deployment
Basically, I'll need to deploy in a first step arcgis server, webadaptor, portal, Image server.
So if i believe
ArcgisServer 10.7 (or 10.8) 4 cores- RAM:8GB
Portal For Arcgis: 2 cores- RAM:8GB
Image Server: 2cores RAM:8GB
And the system should be scalable to incorporate arcgis geoevent server at a later stage
Geoevent server 2 cores- RAM:8GB
In terms of storage I'll start with a 2*2To raid 0 mounted
As a minimum in development I should require 10 cores-32GB RAM 2*2To drives.System should be scalable to 64GB RAM. Physical storage does not need to be scalable yet.
I do ignore the caches needed.
I do ignore connectivity requirements.
However I found such a configuration simply overkill at a first glance. Am I right? Or is it not engouh?
Solved! Go to Solution.
The ESRI maintained, but unfortunately lesser known, "System Design Strategies" WikiGIS pages are your best friend:
System Design Strategies - GIS Wiki | The GIS Encyclopedia
Especially see the:
- Platform Performance - GIS Wiki | The GIS Encyclopedia
- Server Software Performance - GIS Wiki | The GIS Encyclopedia
- Software Performance - GIS Wiki | The GIS Encyclopedia
pages.
To be honest, I personally consider HDD utterly dead for any kind of serious spatial database / GIS related work, especially if it involves any form of (geo-)processing requiring random access to database records, rather than just serving out pre-rendered tiles or so. With the ever growing datasets nowadays, you are going to hit a wall. At a minimum, get SATA SSD, but even better some enterprise class NVMe SSD if you can afford it. I've pounded a cheap 2TB Samsung EVO SSD drive over the past two years writing hundreds of GBs of data to it on a regular basis without issues. Any HDD would probably have failed in the same period under the same workload.
I would also recommend 10 Gbit/s network between the server and any power workstation used for processing or massive data loading. For less demanding purposes, 1 Gbit/s should suffice, if the network bandwidth isn't shared by dozens or hundreds of other users. Separating the power users on a separate 10 Gbit/s sub network from ordinary users on a 1 Gbit/s network, might also help.
The ESRI maintained, but unfortunately lesser known, "System Design Strategies" WikiGIS pages are your best friend:
System Design Strategies - GIS Wiki | The GIS Encyclopedia
Especially see the:
- Platform Performance - GIS Wiki | The GIS Encyclopedia
- Server Software Performance - GIS Wiki | The GIS Encyclopedia
- Software Performance - GIS Wiki | The GIS Encyclopedia
pages.
To be honest, I personally consider HDD utterly dead for any kind of serious spatial database / GIS related work, especially if it involves any form of (geo-)processing requiring random access to database records, rather than just serving out pre-rendered tiles or so. With the ever growing datasets nowadays, you are going to hit a wall. At a minimum, get SATA SSD, but even better some enterprise class NVMe SSD if you can afford it. I've pounded a cheap 2TB Samsung EVO SSD drive over the past two years writing hundreds of GBs of data to it on a regular basis without issues. Any HDD would probably have failed in the same period under the same workload.
I would also recommend 10 Gbit/s network between the server and any power workstation used for processing or massive data loading. For less demanding purposes, 1 Gbit/s should suffice, if the network bandwidth isn't shared by dozens or hundreds of other users. Separating the power users on a separate 10 Gbit/s sub network from ordinary users on a 1 Gbit/s network, might also help.
Many thanks for your insights.
I'll dig the provided documentation.