[ixpmanager] hardware specification recommended

[Nick Hilliard (INEX)]

Hi Basil,

We would recommend 2x route servers and 1x route collector.

The route servers provide bilateral peering, and we view this as part of 
the IXP core infrastructure, so we run two of these.  If you have two, 
it means you can run maintenance on one while keeping the other in 
service.  I.e. the overall peering service will be more reliable.

The route collector is for debugging, and can be used by organisations 
who don't want to connect to the route servers.

We would also recommend 2x RPKI servers for the same reason.

You should have a quarantine route collector to test new connections 
coming into the IXP.

The sflow VM will need fast disks.  SSD is best.  This will only become 
a problem if you have lots of participants at the IXP, e.g. 50+

The RPKI VM scale has nothing to do with the size of the IXP, so you're 
best to ask the Routinator people about that.

INEX uses approximately the following configuration:


	

*cpu*

	

*memory*

rs1

	

4

	

8G

rs2

	

4

	

8G

rc1

	

2

	

2G

quarantine

	

1

	

2G

monitor

	

4

	

4G

sflow

	

2

	

4G

ixpmanager

	

2

	

4G

database

	

4

	

8G

rpki01

	

2

	

2G

rpki02

	

2

	

2G


The physical hardware we run is 2x Dell r730 (Intel E5-2630v4, 10core 
servers, 96G memory).

These VMs are mostly idle. The Hypervisor CPU load is usually around 
10%.  We run the route servers on physically separate hardware (Dell 
R320), just in case we have a problem with the primary management 
hypervisors.

The important thing is that everyone's configuration will be different. 
These are all VMs, so if you need to add or remove memory / CPUs, that 
can be done very easily.

Nick

> Basil Elbalaawi <mailto:saifbasilyazan at gmail.com>
> 5 November 2020 at 16:59
>
> Dear Support
>
> I have almost finished testing the ixp manager full stack system, and 
> will test the sflow next week, so if you can help me to be selected 
> the hardware specification for every service supported to the system 
> Specially about memory, CPU core numbers, storage size for every 
> service? , can you note the following services separated to multiple 
> virtual machine like:
>
> VM1: ixpmanager web front-end VM , and database VM
>
> VM2: monitoring VM (mrtg / nagios)
>
> VM3: sflow VM
>
> VM4: 1 rpki VMs (Routinator 3000),
>
> VM5: 1 route servers with Birdeys.
>
> My server is “HPE ProLiant DL360 Gen10 8SFF” with Esxi hypervisor, all 
> our traffic internal about PSIX- Ramallah with 16 members is from 5-10 
> Gbps.
>
> 	
>
> Memory size
>
> 	
>
> CPU core numbers
>
> 	
>
> storage size
>
> VM1
>
> 	
>
> 	
>
> 	
>
> VM2
>
> 	
>
> 	
>
> 	
>
> VM3
>
> 	
>
> 	
>
> 	
>
> VM4
>
> 	
>
> 	
>
> 	
>
> VM5
>
> 	
>
> 	
>
> 	
>
> Thanks advanced,,,,,
>
>
> <http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail> 
> 	Virus-free. www.avg.com 
> <http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail> 
>
>
>
>
> _______________________________________________
> INEX IXP Manager mailing list
> ixpmanager at inex.ie
> Unsubscribe or change options here: 
> https://www.inex.ie/mailman/listinfo/ixpmanager

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.inex.ie/pipermail/ixpmanager/attachments/20201106/868997f8/attachment-0001.htm>