NCM Web interface is slow!!

So I finally upgraded Cirrus to 5.1 and the web browser interface is still so very slow with no speed improvements anywhere for us. The new features added to the website are great but they take so long to use they are currently useless. On the Node list when we try to view the config files on a node it takes nearly 2 minutes just to show the available configs for one specific device.

We only store the 2 most recent configs per device and we don't have node monitoring enabled. We definitely need some help on this issue.

Has anyone else had similar issues with the website?

Find more posts tagged with

cirrus

website

config

slow

5.1

web

Accepted answers

All comments

chris.lapoint

Ciag, I haven't seen any responses for other users experiencing this issue so it may be specific to your environment. Did you ever get to the bottom of this issue? If not, I think your best bet is to open a ticket with support.

Ciag

Hi Chris,

I agree it definitely seems that way, I haven't resolved this issue yet. I went ahead and raised a ticket yesterday. I will publish any findings after the ticket is hopefully resolved.

juan_kaliente

Ciaran, doing a new install with NCM 5.1. with 6,000+ devices. Have had no issues with the web response, and multiple configs. We are running high-end HP servers.. with the SQL DB separate from the NCM app server on the same Vlan(layer-2), Gig-E speed. So look at your network, and server setup for max performance.

jedi755

We have 4,000 nodes and the web was pretty slow on 5.1, but after installing 5.1.1 it is much faster than before.

My biggest issue was the thwack.com 'widget' on the website slowing performance down dramatically. Removing this option really improved things for us, especially since that server has no connectivity to the internet.

Ciag

Hi,

Apologies for letting this one go a while. This topic has taken a back seat due to more pressing issues we're having with Cirrus. The speed of the web console has definitely increased with the latest release but is still not usable and no where near as quick as our Orion NPM web console which is lightening fast. The starting/loading of IIS is still very slow (first time you open the site or after a prolonged period when IIS times out and needs to be loaded again)

I would love to integrate our Cirrus and Orion Website but would not be willing to until The NCM site is work correctly. Thanks for everyone's comments so far

byrona

I also find the NCM web interface to be pretty slow as compared to the NPM website and in my environment both are being ran from the same server with their database on a separate system.

savell

We also have suffered from slow NCM web response times, particularly when performing an initial logon.

This logon until recently had been taking 2 ½ minutes to complete, which to be honest hasn't made the web interface as popular as it should be.

As a result however of investigations into the NCM orphaned records and subsequent inventory improvements, I also noticed another side benefit - our logon time had changed from 2 ½ minutes to somewhere closer to 1 minute.

Upon further investigation, I found that as part of the logon process, the Policy Violations panel is populated via 4 separately executed queries against the PolicyResultsCache table.

Without any index, each query was taking 25 seconds to execute via a table scan (remember our PolicyResultsCache table has 5.5 million rows).

The query:

SELECT [Table1].[PolicyID] AS _PROJ1, [Table1].[PolicyName] AS _PROJ2, [Table1].[RuleID] AS _PROJ3, [Table1].[RuleName] AS _PROJ4, [Table1].[NodeID] AS _PROJ5, [Table1].[ReportID] AS _PROJ6, [Table1].[ReportName] AS _PROJ7, [Table1].[ErrorLevel] AS _PROJ8, [Table1].[IsViolation] AS _PROJ9 FROM [dbo].[PolicyCacheResults] AS Table1 WHERE (([Table1].[CacheID] = '78') AND ([Table1].[ReportID] = '{394A0E07-6209-4E92-9593-8C71CA351C2C}'))

With the original index I had placed on the table, these queries were taking 400ms each.

A quick modification of the index (nothing flash here as I don't obviously know the table uniqueness rules) brought this down to <100ms per query.

CREATE NONCLUSTERED INDEX [tmp_idx_cacheid_reportid] ON [dbo].[PolicyCacheResults]

(

[CacheID] ASC,

[ReportID] ASC

)WITH (PAD_INDEX = OFF,

STATISTICS_NORECOMPUTE = OFF,

SORT_IN_TEMPDB = OFF,

IGNORE_DUP_KEY = OFF,

DROP_EXISTING = OFF,

ONLINE = OFF,

ALLOW_ROW_LOCKS = ON,

ALLOW_PAGE_LOCKS = ON)

ON [PRIMARY]

Our web site logon now takes <30 seconds, and while still fairly slow is a vast improvement.

I think part of the remaining problem is the number of rows returned by each query. There is a row returned for each policy check which the application code obviously summarises into the displayed table. Some of these initial queries in our environment return >7000 rows to the application for processing. You can see in the SQL profiler trace that it takes a while for each returned rowset to be processed before the next query is issued (i.e. it is a sequential process).

I will log a support case and see if this can be improved.

Dave.

Ciag

Cheers Dave,

that's a fair bit of work you put into finding the solution. The Solution for myself was quite simple in the end. I reinstalled IIS and it seemed to the the 'fix-all'. The homepage loads up very quick now. Unfortunately because of the way the webconsole is designed to run with AD it posses a security risk so now that's my current obsticle. It never ends.

Thanks again for you reply