Measure Your Network Latency Before It Becomes a Problem

The speed of a network is mostly defined every bit throughput: how many megabits or gigabits your network can pump through its pipes in the course of a second. That'due south certainly important, but at that place's another factor to consider, and these days it'south often at least as important (and sometimes more important) than raw speed. That factor is latency. The best way to define latency is equally lag: the span of time betwixt when a bundle or group of packets is sent from a source to a destination, and it's usually measured in milliseconds (ms). To users, even big differences in lag time aren't ofttimes noticeable in the course of using their everyday applications. That is, unless they're using latency-sensitive apps. The list of latency-sensitive apps has been growing lately, which is why latency is so important. Most apps tin can adjust for a slower latency, but some—peculiarly those that rely on video streaming or some other kind of constant data feed—tin balk if some minimum latency performance isn't met.

A fundamental driver that's irresolute the mural of latency is the deject service. Cloud services, whether they're running in private data centers or public clouds, such as Amazon Web Services (AWS) or Google Cloud, are by necessity located at a distance from your data eye. Normally, the path to those cloud services travels over the public internet. For most network traffic, an internet connection is fine. But for latency-sensitive apps, the net is an unpredictable wilderness that tin exist calm one millisecond and stormy the next. This is because the data path is subject to the vagaries of routing and processing delays. These are factors that neither your It staff nor the deject service'due south IT professionals can command.

What Is Latency?

In that location are two types of latency that affect cloud services. I is the fourth dimension it takes data to travel across the network from one signal to another. This is limited by the "wire speed" of the connection, meaning either the speed of light as it runs through a drinking glass fiber-optic cable or the bespeak speed of information running along a True cat X copper cable. In all cases, information technology's directly related to the altitude involved. Most important, there's goose egg y'all can do about the delays caused by the time it takes for the transmission, except to move closer to the cloud provider or to choose the closest server location for the cloud service y'all're using, though not all deject services give you that pick.

The second blazon of latency is created past processing delays or sometimes past poor routing. Each time your information packet passes through a device, at that place's a brief delay while the router or switch decides where to ship it next. There are additional delays caused by packet inspection in firewalls or other security devices, and by servers as they receive a request for information, detect it in their mass storage, act on the request, and transport a response. While each of those delays may be cursory, they add together upward. Sometimes, if your path from your data eye to your data in the deject is congested, so the delays, and thus the latency, can really add up.

Measure Latency With Traceroute

Fortunately, measuring the latency for your data is adequately easy, and it doesn't price anything. To find out, run the command line in the operating organization (Os) of your choice, and perform a Traceroute control. In Microsoft Windows, you type in the command "tracert" at the command prompt, followed past the proper noun of the destination, perhaps "aws.amazon.com" or "cloud.google.com." (See screenshot below.)

In Apple OS X or Linux, jump to the terminal window and type: traceroute -I companydomain.com. Alternatively, yous tin can replace the domain proper noun with an actual IP accost. You want the -I switch so that traceroute will utilize the Internet Control Message Protocol (ICMP), which it's going to need in gild to accurately mensurate latency.

No matter the OS, what you run across on the screen are responses from each router along the pathway to the website you're checking, along with a time measurement in ms. Add those upwards and that'due south your latency. If you run the command several times, and then y'all'll notice the same routers don't always testify up. This is considering your data packets may take unlike internet pathways each time y'all run the command. Some pathways can be longer than others, adding to your overall latency. Welcome to latency on the net.

Click screenshot to enlarge.

Measure out Latency With Network Management Tools

IT pros in charge of a corporate network volition have some kind of network monitoring tool to proceed tabs on what's running beyond their wires and how that traffic is doing. You can check network latency in other ways likewise. If it's a well-featured tool, then it will have its own latency measurement features. For example, Spiceworks has its own Traceroute software that uses Manual Control Protocol (TCP) rather than the ICMP protocol used in the standard traceroute role outlined earlier. That'southward also used in the Ping command. In some cases, TCP can be more than accurate because some routers are configured not to respond to ICMP packets.

Some other adept example is Paessler AG, which sells the Paessler PRTG Network Monitor. This tool will provide network latency along with its other functions. Paessler AG has a freeware version of its software also as a free trial. I will be looking at the PRTG Network Monitor, including its latency capabilities, as I test the tool for an upcoming PCMag review.

"Beyond just ping times, if I'g measuring that every menses of time, that'south the basic measure of latency," explained Greg Ross, Senior Systems Engineer at Paessler AG. "You tin break up the path, and evaluate for each hop in the path. We can do that with a trace road hop count.

"The other attribute is Quality of Service [QoS] monitoring, which looks at jitter and roundtrip fourth dimension between me and the endpoint," Ross said. He said that such monitoring tin also show other causes of latency such as packets delayed or packets received out of order.

Reduce Latency's Bear upon

If your apps are having trouble with high latency times, then there are a few things that you tin do to convalesce latency's impact, even if you can't eliminate it. The start is to investigate which Domain Name System (DNS) service y'all're using if you lot don't have your own DNS server. I hidden crusade of latency is the filibuster that occurs while the DNS server you're using looks upward the internet protocol (IP) address of the website you want to attain.

A distant DNS server will introduce latency equally will ane that simply doesn't perform well. Having your own server will reduce that lookup time, provided the accost is in your server's tables. Otherwise, you'll accept to await while it asks the next DNS server up the line for the address. For websites you lot visit frequently, this tin cut latency.

Having a dedicated connection will also reduce latency, provided it's really dedicated, pregnant you have a connection using a defined line. This can be a concrete fiber connection if you're close enough or a line leased from a carrier. This fashion, y'all reduce the number of routers involved, and you reduce the take chances of routing errors that tin cause latency.

And, of course, you can reduce the distance. According to tables provided past M2 Eyes, 100 KM of fiber introduces nearly 500 microseconds of latency. That's a half millisecond, then you run across how distance tin can make latency add together upwardly.

Keep That Hop Count Small

No matter what "cure" you use to deal with latency, be sure to consider how information technology will impact other aspects of your network. For example, if you're using your cloud for disaster recovery (DR), then moving besides close tin can crusade its ain set of bug since you don't want the aforementioned disaster to hitting your DR site as hits your data eye.

In general, yous can reduce latency past reducing the number of hops in your pathway, and there are several ways to approach that trouble. Aside from those mentioned before, you lot can take guaranteed bandwidth to reduce latency acquired by congestion, or you can have redundant providers so that your app can cull the path with the least latency in real time.

While you can't eliminate latency without violating the laws of physics, you also tin can't ignore information technology. With popular apps such as existent-time streaming analytics, video conferencing, and fifty-fifty your company's Voice-over-IP (VoIP) traffic all becoming ever more latency-sensitive, digging into your network's latency now, before information technology becomes a problem, will pay dividends down the line.

Source: https://sea.pcmag.com/ipswitch-whatsup-gold/28764/measure-your-network-latency-before-it-becomes-a-problem

Posted by: radleythemblent.blogspot.com

Share this post