I currently work in an environment that is using a 7 server SolarWinds implementation (NPM, APM, NCM, Netflow, IPSLA, Kiwi Syslog, Storage Profiler) to monitor a range of Cisco Network equipment; HP (Proliant, Blade, DS, GS, RX, EVA SAN), IBM (X, Blade, P570, DS4xxx SAN, DS8xxx SAN, N Series NAS) and Dell (PE) Hardware running many variations of Windows, ESX, Citrix, Novell, HP-UX, HP-Tru64, HP-VMS, Linux, AIX and also a range of ancillary security and other devices such as UPS/EMU/PDU/AirCon Units. At present we are monitoring a core environment of ~1700 devices but this is quite likely to expand considerably.
UDPs are a very necessary capability to cover this range of equipment and the display of data exposed through UDPs is very important as we want to provide a reasonable level of basic data and alerting before more focussed troubleshooting may be undertaken using products such as HP SIM, IBM Director, Dell OpenManage, SCOM and various storage and other vendor utilities. With the range of equipment we are monitoring I am having some difficulties with the organization, naming and sorting of UDPs as well as with with the display of UDP data in views and would appreciate any feedback that can be provided. I currently have ~270 UDPs configured.
I've looked through Thwack and have seen some of my issues raised but not with a lot of detailed discussion about how to deploy UDPs in a scalable manner, how others have dealt with naming and display ordering and what is coming out real soon now and what is in the longer pipeline for future releases.
1. When defining a UDP it appears that it must be unique throughout the environment even though it can be grouped into a single level of groups. This means that a structured naming convention needs to be developed to allow for things like serial number UDPs for different OIDs for HP, Dell, Cisco nodes etc even though they are grouped by different groups such as HP Server, HP Blade, IBM Blade, Dell Server, Cisco etc. It also appears that a UDP Name cannot start with a number and may also need to incorporate display ordering characteristics (see 2). The UDP Name will also be the default column heading in some resources as well as being used for the default chart heading in another resource. Given the above the naming convention becomes very important.
2. When displaying UDPs in a UDP Summary Status resource it appears that the display order of these UDPs is alphabetic only as there is no other control that I have been able to find. This means that the structured naming convention also needs to allow for display ordering as well. I note that the Group Name appears in the resource display in RED and that multiple Group Names are catered for and automatically appear as Sub-headings within the resource.
3. When creating UDPs to expose SNMP tables for display in a Tabular UDP resource, if a UDP Summary Status resource is in play then these UDPs display quite useless information in the Summary Status resource showing the number of values (table rows), and apparently can't be turned off. With many Views and Resources already being quite busy it would be good to be able to remove useless or unwanted information.
4. When displaying UDPs for SNMP tables in a Tabular UDP resource the length of the UDP Name is important as it is used by default for the column label (the label can be different but this means extra work). If the labels (or names) are too long and the heading goes to two lines then a strange scroll bar scenario comes into play. It also appears that there is no way to control the order of the columns, which appears to be random, nor the justification of the column data.
5. When gathering information about a range of devices (say Blade Enclosures or UPSs) it is really useful to be able to display a UDP for a group of nodes using a Node Details resource, but it appears that this can only be done for one UDP at a time. As such multiple Node Details resources must be used in a view to display multiple UDPs (clunky) for the same group of nodes, where a single resource displaying multiple UDPs in a multi-column tabular format (eg. Node, Model, Firmware, Status in columns) for the group of nodes would be much more valuable and usable in many situations (end users find the need for multiple resource display not particularly elegant, and I agree).
6. When selecting the UDP to be displayed in the range of UDP resources the drop down provides an unsorted list of all UDPs without any grouping being applied either. This is not particularly useful when a large number of UDPs are involved. The lack of grouping may add another dimension to the UDP naming convention.
Thanks in anticipation for any feedback or examples.
