2014-03-23 16:24:00 UTC

When evaluating Network Performance Monitoring, what aspect do you think is the most important to look for?


Let the community know what you think. Share your opinions now!
Guest
4646 Answers

author avatar
Top 10User

Expectation setting and network segment focus are two of the most important actions that I look to take when looking at network performance and optimization. As an ex-Cisco TAC agent, "the network is slow" was the constant drone on the phone from many a customer. "The network works, no excuses" was the company mantra, in the 1990-2000 decades when I was there. But, the reality was the most of the solutions were a matter of discussing anecdotal observations with the Customer while mapping-out a crude topology. Rarely did we get a Customer with a topology map. We did this so often, that most of us bought magnetic drawing pads, at Toys'R'Us, to quickly draw and erase topologies from case-to-case. It sure saved on paper and ink.

1.) Are my users trained enough to understand that a slow network can be the internal corporate network OR the Internet OR both?
2.) Can my users understand that performance is anchored on the LAN or WLAN separately and at the junctures of these two segments?
3.) Do my users understand that an overcrowded ZOOM site, on the Internet, may be killing their VTC and NOT the company networks?

Then, there are the financial questions;

1.) Can I find a tool that will let me see clearly what my LAN, WLAN, and WAN? At a great price and with great support?
2.) Should I hire someone who knows how to deal with the integration of OpenSource tools that have been used for years?
3.) I have a Windows environment with some MacOS. Why should I have to learn and set up a *NIX system just to run OpenSource tools?
4.) I heard PRTG r-o-c-k-s, BUT, it may be overkill for my network. Sophos, Sonicwall, ConnectWise Automate, Auvik and other such tools may be all I need.
5.) I've got new 10Ge and 40Ge segments that came with my SAN. Can these tools adjust to analysis on these higher speeds?

My suggestion;

First, make sure you empower your users with a tool that will help them figure out where the slowness is occurring. I would highly suggest contacting ookla.com, the speedtest.net folks, and look into there corporate product. You have iPerf agents that they customize for use on Android, iPhone, Windows, MacOS, *NIX, etc. Your users use the app or internal website to run tests to determine whether the LAN or WLAN are having issues OR whether its the internet. The apps can be customized for company logo, support links, etc. as much as the web site format. It stopped a lot of unnecessary user complaints to the Help Desk.

Once you got your users set at the right expectations and empowered with an effective tool, you can work on finding the omnipresent, inexpensive and well-supported tool or tools that can help you make sense of what your network segments are doing and how to tune/optimize, thereafter, should that be required.

2019-08-07 00:17:03 UTC
author avatar
Consultant

1. able to monitor the CPU, Memory, Response Time, latency, packet loss, Interfaces traffic, and Bandwidth utilization
2. able to trace the connectivity of each devices
3. capability to create an alert
4. capability to generate a report
5. all or most of the devices should be supported.
6. Must be a user-friendly

2016-01-04 05:53:21 UTC
author avatar
Vendor

Cost, how many feature supported by the tools and complexity to maintain

2015-05-11 07:35:10 UTC
author avatar
Top 20Real User

Historical graphical display of all the common metrics(bandwidth, latency, jitter, etc.) that should be monitored on an network.

2017-01-31 20:08:21 UTC
author avatar
Vendor

Network performance, monitoring and reporting software shoud have (visible) feature set suited for the purpose. Users must feel themselves confident to use it. Intuitivity. Stability and reliability as well as speed is a must (appropriate application code language). Also I am looking for a highly configurable alerting.

2015-08-22 15:51:15 UTC
author avatar
Vendor

Typically I'm looking for latency, packet loss, total bandwidth utilization, QoS and jitter.

2015-08-13 21:57:51 UTC
author avatar
Real User

The most important criteria is the monitoring tool must be vendor neutral to cover most networking brands out there, and should use open standards like snmp. It should not require additional software agents to be installed on the nodes to be monitored

2015-05-06 22:12:31 UTC
author avatar
Consultant

There are so many products to choose from on the market, and each has its strengths and weaknesses, so I would probably say that the most important criteria is "fit for purpose". So for example, in an enterprise environment I would be looking at scalability as a major factor, with network bandwidth usage also a factor if monitoring across a wide area. If I wanted to measure latency, then I would want a product that can do this with an accuracy around +-10ns across multiple technologies - servers, switches, routers etc.
Probably the most important general criteria would be TCO, which is really a combination of many of the responses above. TCO factors would include purchase cost, implementation cost, ongoing management costs, requirement for and availability of specialist staff, ease of use and availability plus cost of extensions.
These are probably my main criteria. In the real world, I have a spreadsheet with over 100 different evaluation criteria, which would be supplemented by another sheet detailing the specific requirements of the customer.

2015-04-21 11:23:40 UTC
author avatar
Vendor

Packet loss, bandwidth, easy reporting, being able to filter information, select different criteria.

2015-01-06 18:02:32 UTC
author avatar
Consultant

the ability to have everything in one location. I want to be able to see what is all going on with just one click and not having to search through multiple pages to find one thing i need

2014-11-06 20:28:12 UTC
author avatar
Vendor

Behavior of an application or service provided to end users/customers with the ability to measure and improve upon service delivery

2014-05-07 16:39:59 UTC
author avatar
Real User

Since monitoring is the key topic of this question, the reliability of the software is the single most important aspect to consider. No matter how clever of how easy the solution is to use, if the system isn't reliable that is in direct conflict with its purpose. First look for reliability then prioritize your key requirements.

Monitoring is a very competitive field and a great many solutions are available and the top picks from this forum are all excellent.

I have found in my decades of experience in this field, that the biggest challenge in the reliability of monitoring systems is its *reliability under scale* and scale usually is the most problematic when you need it most... when the network has widespread outages. This condition puts a monitoring system under tremendous strain as there are great many network conversations that are simply timing out and for conversation that is waiting, that's time the system is loosing and not available for actively monitoring. There will be processes or callbacks sitting in a queue that will stack up that eventually will block the system's ability to do other work gathering performance information from the rest of the network. This usually happens when there are 10s of thousands of devices and hundreds of thousands of resources and millions of metrics to gather in a short period of time.

A performance monitoring system (any monitoring function actually) that scales, in general terms, must be highly asynchronous. This adds massive complexity to the software, especially when the system is distributed, and takes a very long while to work the majority of issues out of it because that can only come from experience. Fortunately, open source projects can compress this time immensely due to the number of users so, when selecting a solution that is open source, the age of the project is important when it comes to reliability. The quality of the open source solution can also be evaluated by how quickly issues are addressed and resolved by the development community.

In summary, first evaluate for a system's reliability when it comes to performance monitoring. This can be measured by how well the system handles scale, scale under stress, and then, when open source, evaluate the health of the project by its release cycle and responsiveness to bugs; which is story that can only be told over time. Any system that has been around for more than a couple of years will certainly have the basics to meet your needs and if it's open source, you have the opportunity to provide feedback and request features the you believe it needs to meet your requirements.

My $0.02.

2020-01-27 13:53:03 UTC
author avatar
Vendor

Usability.

2019-08-30 18:27:03 UTC
author avatar
Top 5Real User

There are many solutions for the monitoring system.

However, install and provide solutions with cacti under Ubuntu server 15.04

Ideal for large and small networks under snmp cisco among others

2019-03-24 16:17:42 UTC
author avatar
Real User

Monitoring interfaces and services. It's very important for me to monitor certain services.

2019-03-22 16:27:48 UTC
author avatar
Real User

I look out for nodes that go off frequently especially when you have secondary providers.
I look out for nodes maxing out on the bandwidth for planning and budgetary purposes
Bandwidth hoggers

2018-12-05 13:04:33 UTC
author avatar
Top 5LeaderboardReal User

- Highly detailed stats collection without roll-ups
- Out-of-the-box Monitoring and alerting
- Product has to be multi-user and multi-vendor
- As little system maintenance as possible
- Reliability and Scalability
- Customization
- Fast development cycle
- Good support

2017-12-09 04:38:12 UTC
author avatar
Consultant

Working for NETSAS Australia, who developed Enigma NMS, we have learned from our clients that they need:

- All features in one product, they don't want to buy multiple products and then go through the integration troubles.
- Highly detailed stats collection without roll-ups
- Out-of-the-box Monitoring and alerting
- Product has to be multi-user and multi-vendor
- As little system maintenance as possible
- Reliability and Scalability
- Customization
- Fast development cycle
- Good support

However, regardless of great features and scalability Enigma NMS lacks flashy GUI and it is quite complex primarily engineering tool. It takes a bit of time to learn how to use it.

2017-10-23 01:00:08 UTC
author avatar
Consultant

Features, how many different products or sub-products you need to buy to cover all or most of your operational and management needs.
Product integration, how much will it cost you in HR to ensure that all products within your network management and monitoring solution act as a single system. They all need to access the same database of nodes, sites, vlans, MAC, contacts, etc.
Scalability, how many server instances you need to have to monitor everything you need to monitor.
Statistical Data resolution, data averages and roll-ups are bad.
Maintenance, how difficult is to maintain the product. Do you need to hire a dedicated person to maintain the solution.
Analytics and Automation, how easy to extract required information for vast volumes of data.
Integration with 3rd party products and REST API.

2017-10-23 00:49:25 UTC
author avatar
Vendor

Consistent feedback from our customers include these top 5 capabilities;

1. see the real-time status of the network by summary, site, device, interface and application
2. proactively receive alerts such as threshold crossing events, path changes, etc
3. the ability to drill down from site to interface to diagnose and fix configuration or policies relating to QoS effecting app performance/user experience issues eg. pkt, loss, jitter, latency, etc
4. having historic play-back of raw flow data for forensic analysis eg. 48 hours, 2 weeks, 30 days, ...
5. dashboards and reports for governance eg. SLAs

2017-10-10 16:48:50 UTC
author avatar
Vendor

Most of the world class NPM like Solarwinds/WhatsupG/PRTG are similar in respect to ability to process in real time but I think the most important should be Notification/Alert (how quickly user can get notified), second Interactive Dashboard and third Reporting...( Which I found lacking in most of the above product I mentioned.) ....Hope its help....

2017-07-12 13:01:32 UTC
author avatar
Vendor

Overall, I feel you need to look for continuity, compatibility, and consistency. SolarWinds provides this, it also provides customizable interfaces to fit everyone's needs. With a always updating MIB database and several ways to identify hardware, NPM is the best tool I have ever used for Network monitoring.

2017-07-05 13:33:53 UTC
author avatar
Vendor

Ease of adding end devices to be monitored by the system.
Intuitive and lucid interface.
Ease of administration and maintenance.
Extensive monitoring points (rather than typical up/down and CPU/memory/disk).
Flexibility to develop customization (SAM allows to create templates, NCM allows to run cross platform scripts, etc.)
Quick and responsive interface.

2017-05-02 18:40:14 UTC
author avatar
Top 5LeaderboardUser

Historical data, trends, and up time. Understanding the metrics involved, especially in network monitoring will lead to a better forecast of improvements.

2017-05-01 12:26:05 UTC
author avatar
Consultant

i use zabbix nms for snmp and wmi monitoring and graylog for syslog message analysis, also i monitor graylog from zabbix using the api (a simple script based on curl and jq)
actually i monitor more than 1300 devices and zabbix feature like remote proxy, graphical maps and trigger dependancy save my life tons of times.
Also with graylog i took the not gelf format log and i send them to ossec server and get back via api possible security warnings.

2015-12-09 16:28:15 UTC
author avatar
Consultant

How fast the netwrork is
East to use and installation
Performance
Monitoring and reporting software with logs
If error displays how to figure out to resolve the issues faster

2015-11-18 13:37:34 UTC
author avatar
Vendor

Would like to know 1) End to End, 2) Multi Vendor environment, 3) Easy to Use and Installation and 4) Analytic Reports..

2015-10-05 05:04:00 UTC
author avatar
Consultant

Real data, multi vendor, clientless, easy to use with an easy to understand dashboard

2015-09-29 15:06:09 UTC
author avatar
Vendor

must use actual data from real client sessions to see users true experience. i.e. not use 'artificial' data that is sent to test the link

2015-09-29 13:53:52 UTC
author avatar
Real User

In sequence of importance Latency , packet loss,server response times

2015-08-08 12:01:28 UTC
author avatar
Vendor

Working for Paessler, a vendor of a network monitoring solution I can only give an idea of what customer's main criteria is due to my experience with trade shows and other events, and that's usability, usability, usability and beyond that a feature set that fits to your needs - that can be very different. And of course also licensing and price are important aspects when coming to a final decision.

2015-06-24 13:03:37 UTC
author avatar
Vendor

Packet loss, bandwidth monitoring, memory, CPU usage, nodes monitoring, packet capturing, reporting and alerts.

2015-05-11 10:04:43 UTC
author avatar
Vendor

Bandwidth usage by various type of transport protocols

2015-04-28 13:47:25 UTC
author avatar
Consultant

Packet loss, bandwidth specific to VoIP, ease of use, TCO

2015-04-28 12:36:00 UTC
author avatar
Vendor

Prevention, identification and correction of irregularities such as slowness, inaccessibility and undesired interruptions of the IT infrastructure

2015-04-23 12:43:00 UTC
author avatar
Vendor

Precise Root cause Analysis with a Network Mapping!

2015-03-05 13:56:18 UTC
author avatar
Real User

I start with ease of use and ease of implementation followed by how it works with different vendors and devices from UPSs to servers to switches

2015-02-12 15:30:21 UTC
author avatar
Vendor

Vendor inter-operability is a big consideration. Network monitoring should also focus solely on the network and avoid application monitoring. Products that try to do it all under one of pane of glass tend to have important features missing or not available. Within the network monitoring application, I would like to see all network performance metrics and perform packet captures and analysis - if at all possible. The program should use a variety of industry protocols for monitoring and vendor-specific monitoring protocols like NetFLow, J-Flow and others. The ability to customize the monitoring and presentation of the data are key but the program should have an easy to view default dashboard. If it's too complex to implement then I would move on to another product.

2015-01-28 16:34:38 UTC
author avatar
Vendor

Packet Loss while transferring different size data across the network.

2015-01-21 20:13:13 UTC
author avatar
Vendor

When monitoring network performance mainly we need to find out from which node packets are high and find out whcih process us utilizying it.

2015-01-03 18:11:16 UTC
author avatar
Vendor

First of all, reliability and accuracy. If I can be sure that the measurment is trustworthy, I find if the features covers what I´m looking for.

2014-12-01 18:53:08 UTC
author avatar
Vendor

Ability to process in real time without having to store packet capture. I only want to store meta data and us etriggers for packet capture.

2014-10-22 04:57:10 UTC
author avatar
Consultant

I think that most important the posibility of clear and simple views and granularity, is diffcult expense time to understand a lot of graphs, in my case on Vmware tools like vCOPS (vCenter Operations Managers) is very easy start in a simple view and make go to deep troubleshoot and reports.

2014-10-02 15:36:43 UTC
author avatar
Vendor

From the perspective of network administration, I look at things like packet loss, latency, jitter, downtime, interoperability with the largest amount of vendor devices possible, performance and behaviour of the application.

2014-06-28 14:55:37 UTC
author avatar
Vendor

I tend to look at network performance from a VoIP perspective, so I'm looking at things like latency, packet loss, jitter, and end-to-end QoS.

2014-06-22 17:20:17 UTC
author avatar
Vendor

When monitoring network performance I look at the process, disk system activity, memory and cache usage. Monitoring legacy applications.

2014-05-15 16:27:38 UTC
Find out what your peers are saying about Paessler AG, Zabbix, Centreon and others in Network Monitoring Software. Updated: May 2020.
418,350 professionals have used our research since 2012.