ACE health monitoring queries

[nsyal]

Hi,

I need some detailed information regarding ACE health monitoring. According to the documentation, ACE support 4096 Unique probe configurations.

My questions are as follows:

1) Are these uniques configs means uniques instances or unique probes. For ex: If I have created 2 probes(say http and icmp probe) and have applied each of it to 5 rservers, does that mean I have 2 unique configs or 10.

2) How did this number(4096) came in. What is the logic behind it. Is it based on the memory allocated by ACE to health monitoring. Please provide a detailed answer to this question.

Any inputs will be highly valuable for me.

Thanks,

Nitesh

[Syed Iftekhar Ahmed]

There are three different types of probe limits in ACE

1.Probe definitions (4K)

These are the actual number of probe objects that you can define.

You can define maximum of 4K unique probes.

2.probe Instances (16K)

Each time that you use the same probe again in a different serverfarm/Real

the ACE counts it as another probe instance. You can allocate a maximum of 16k probe instances.

3.Simultaneous Probe Sockets (2500)

With ACE 2.x code Probes can open 2500 simultaneous connections from the control

plane. In Older ACE module codes and on ACE appliance this limit i 1000.

Reason for these probe definitions & Probe instances are the Control plane capacity.

These limitations exist to make sure that Probes do'nt starve the other process

Hope this helps

Syed Iftekhar Ahmed

[nsyal]

Hi Syed,

Can you please elaborate on the Control Plane Capacity. From the hardware I can see that there is 2G capacity between the CDE switch and the control plane. I want to know how did the number came in. What is the calculation behind it?

Thanks,

Nitesh

[Syed Iftekhar Ahmed · Accepted solution]

In ACE Control and data path runs on seperate processors.

A seperate processor for control traffic ensures that heavy Load balancing

connections have no impact on the control traffic.

Processor responsible for Control path (along with Health montitoring)

is responsible for syslog, snmp, ARP, High availabilty and device control.

As I wrote earlier if unlimited probes are allowed on ACE then there is a possibility

that probe processing will choke the CP Processor.

These probe limits do not guarantee that you will be able to run these many (16k) probes.

There are so many factors involved. Different probe types have different performance impact.

Probes that parses Layer 7 traffic are definitely heavy on processor than Layer 3 probes.

Similarly Health monitoring frequency can also put additional stress on CP.

Syed Iftekhar Ahmed

[stephg]

For an ACE module, are these numbers for the module, or for each context? Would there be a way to find out how much each context is using?

Regards,

Stephane