Community
- Command Central
- MVP Program
- Monthly Mission
- Blogs
- Groups
- Events
- Media Vault
Products
- Observability
- Network Management
- Application Management
- IT Security
- IT Service Management
- System Management
- Database Management
Content Exchange
- SolarWinds Platform
- Server & Application Monitor
- Database Performance Analyzer
- Server Configuration Monitor
- Network Performance Monitor
- Network Configuration Manager
- SQL Sentry
- Web Help Desk
Free Tools & Trials
Store

Where do I find latency rather than response for a report?

We've been asked to generate some KPI's for certain bits of kit - specifically for latency. The first bit I can manage, only the problem I have is that I can't latency labelled as latency. So is it masquerading as something else and if so can someone please share?

It could be that response is actually what I need, it just uses the wrong label. What confuses me is that I can add a widget called 'Network Latency' so my presumption is that I must be able to see that as a record in the DB... surely?

N.B. for the record I can find response, response min and response max but as any network bod will tell you "response does NOT equal latency"
p.s. I am a network bod, this SolarWinds lark is a side job I've been lumped with

Find more posts tagged with

Accepted answers

All comments

jm_sysadmin

So generally in Orion shows Average Response time (from the Orion server) as Network Latency. Latency means a bunch of things when you monitor, so its always specific to what you are thinking about. A VMware admin or a storage admin, or a windows admin will likely all pick different metrics.

Now if you have a specific metric that you want, let me know and we can find it. If I were populating a KPI widget, I would want a count of the nodes where Average Latency was higher than some number. For me it's over 20 ms for warning, and 100 ms for critical, I'd show both counts and I'd link it to the resources elsewhere that lists those devices.

Honestly though on our network latency isn't a concern even for our offsite locations. I do care about utilization of links, errors and discards. There are also some SaaS applications that I pay attention to the latency that we pull from NetPath. Its likely not fixable by our staff (we might call a vendor) but that we can give our Support Staff a warning that calls might be inbound, and that its not us but them, and we are working with them.

jm_sysadmin

Oh, and the best network latency (since you are thinking about that) come from IPSLA tests that Cisco can do, and VNMQ can monitor. They obviously are specific to a path between 2 cisco devices and I think typically a TCP port. But they will also give you jitter and other metrics.

stuartd

It's a thorny issue that opens whole cans of worms designed to chew you up and spit you out the other end feeling all confused.

The client wants KPI checks and it is seen (by them) that SolarWinds is the holy grail and should do absolutely everything include getting your shopping and managing your budget.... the trouble is, they refuse to identify what those KPI's are, so in a way it is up to us to decide, document and defend.

We are unclear on what they want, why they want it and what they'd do if a figure doesn't meet their unknown expectations

I guess initially I'd like some clarity and may need to ask SolarWInds directly, what exactly response is measuring, where it measures it from (presumably the local poller), and how it gets to its end figures. I'm working on the basis of getting response (to me response is the round trip time) to their 3rd party routers in to a report. Easy enough done, but then we need to negotiate how often, how do they want the stats presented (raw or pretty), what sort of averages are they looking for!

At this stage, no widgets are involved and my prime reason to mention them, was purely to ask where does that widget get its stats from and what do they mean by latency. Once I know this I can fight the fight, but realise I may need to log a ticket.

In the meantime, I would like to try and extract from the DB, the network latency (if that isn't what SolarWinds mean by response) to any one node or multiples of.

mesverrum

Response/latency in npm are interchangeable terms for NPM. They are the round trip time for an up/down check on that node from the polling engine it is associated with.

Easiest place to grab it is a view

SELECT top 1000 * FROM [SolarWindsOrion].[dbo].[ResponseTime]

Keep in mind that for data that has not been rolled up (< 7 days by default) then min/max/avg will all be the same. Those columns only take effect after a rollup to the hourly and daily averages.

One fun wrinkle that you may run into sometimes is that the response is normally ICMP ping, but can be switched manually on each node to getting response status from SNMP, which will dramatically slow down the response rate as we now have added in the delay of answering an SNMP request.

Setting KPI's can be a bit of a rabbit hole sometimes. With almost every client I worked with the KPI's they wanted were just picked out of a hat to give them something to show at the exec meetings to make them look good. Network latency is basically just a problem for people from the 90's or who are stuck on satellite uplinks, the terrestrial network for the entire globe is pretty fast and reliable these days. Assuming you aren't one of those unlucky souls then the latency number is just going to be one of those easy wins for the report. "Yep, still green"

stuartd

Ahh, as I thought it might be. Thanks for confirming Marc.

[quote userid="245123" url="~/product-forums/the-orion-platform/f/report-lab/91293/where-do-i-find-latency-rather-than-response-for-a-report/290860#290860"]Keep in mind that for data that has not been rolled up (< 7 days by default) then min/max/avg will all be the same. Those columns only take effect after a rollup to the hourly and daily averages.[/quote]

Understood, though I'll potentially forget this bit in the near future.

[quote userid="245123" url="~/product-forums/the-orion-platform/f/report-lab/91293/where-do-i-find-latency-rather-than-response-for-a-report/290860#290860"]"Yep, still green"[/quote]

Yes, I suspect they may be one of those types of client that require this sort of response