When you make the switch to the cloud, you will likely discover that the standard QoS techniques, from the days when services were hosted within your enterprise, will not work on traffic coming in from the public Internet. Below we detail why, and offer some unique alternatives to traditional router-based QoS. Read on to learn about new QoS techniques designed specifically for the Cloud.
Any QoS designed for the Cloud must address incoming traffic not originating on your Network
Most Internet congestion is caused by incoming traffic. From downloads of data not originating at your facility. Unlike the pre-cloud days, your local router cannot give priority to this data because it has no control over the sending server stream. Yes, you can still control the priority of outgoing data, but if recreational traffic coming into your network comes in at the same priority as, let’s say, a cloud based VOIP call, then when your download link is full, all traffic will suffer.
Likely No Help from your service provider
Even if you asked your cloud hosting service to mark their traffic as priority, your public Internet provider likely will not treat ToS bits with any form of priority. Hence, all data coming from the Internet into your router from the outside will hit with equal priority. During peak traffic times, important cloud traffic will not be able to punch through the morass.
Is there any way to give priority to incoming cloud traffic?
Is QoS over the Internet for Cloud traffic possible? The answer is yes, QoS on an Internet link is possible. We have spent the better part of seven years practicing this art form and while it is not rocket science, it does require a philosophical shift in thinking to get your arms around it.
How to give priority to Cloud Traffic
We call it “equalizing,” or behavior-based shaping, and it involves monitoring incoming and outgoing streams on your Internet link. Priority or QoS is nothing more than favoring one stream’s packets over another stream’s. You can accomplish priority QoS on incoming streams by queuing (slowing down) one stream over another without relying on ToS bits.
How do we determine which “streams” to slow down?
It turns out in the real world there are three types of applications that matter:
1 ) Cloud based Business applications. Typically things like data bases, accounting, sales force, educational, Voip services.
2) Recreational traffic such as Netflix, YouTube
3) Downloads and updates
The kicker that we discovered and almost always holds true is that Cloud based applications will use a fraction of the bandwidth of the video recreational traffic and the downloads. If you can simply spot these non essential data hogs by size and slow them down a bit, there will be plenty of room for your Cloud applications during peak periods.
How do we ensure that cloud traffic has priority if we can’t rely on QoS bits?
To be honest, we stumbled upon this technique about 12 years ago. We keep track of all the streams coming into your network with what can best be described as a sniffing device. When we see a large stream of data, we know from experience that it can’t be cloud traffic, as it is too large of a stream. Cloud applications by design are rarely large streams, because if they were, the cloud application would likely be sluggish and not commercially viable. With our sniffing device, the NetEqualizer, we are able to slow down the non-cloud connections by adding in tiny bit of latency, while at the same time allowing the cloud application streams to pass through. The interesting result is that the sending servers (the same ones that ignore TOS bits) will actually sense that their traffic is being delayed in transport and they will back off their sending speeds on their own.
For more information or a demo feel free to contact us http://www.netequalizer.com.
For further reading on this topic, check out this article: “Traffic Management, Vital in the Cloud”