Does the peak look familiar? This is the reality of our daily life and most of the time we simply follow the queue blindly. Each queue system can be questioned for its effectiveness and can be improved and the same principle applies to the queues in our computer systems.
NetScaler keeps so many different queues for different system features and capabilities. The queues admirably and work mostly as an end user you will not even notice when the queue is maintained. Couple return of the month, we reviewed the internal queues maintained in NetScaler and the thought was to improve the management of the queue and handling what we do today.
A simple example of handling client connection on NetScaler. We pick the client connections and based on the LB / persistence logic reuse the selected service. What happens when the service is overloaded or the back-end server are not able to handle more traffic? Then, client connections begin to queue on the service and the queue starts to grow. Most protection functions are used at each level of service today and thus while making the decision LB vserver does not know the status of these features such as HTTP DoS protection, Priority Queuing or Sure Connect. Hence the sometimes selected service may not be the best because of these protection features. All priorities is going to the service layer and the protection logic is applied here so through multiple service prioritization differ depending on waiting queue level of local service.
Another common problem seen with the service level queue when the service is down for any reason any queue is cleaned and connections held triggers. If you combine all this together with the fact that today the logic of the application is mainly applied to the layer / vserver endpoint when you see the opportunity for improvement. Does the story come together ... ??
With NetScaler 10.1 release, we introduced the queue mechanism enhanced with maintaining the queue to the layer of vserver. Again, as an end user you would not see much difference outside as most of it is handled internally, but you'll certainly appreciate the benefits that come
- First and first of all -. One checkpoint
- Play with the size of the queue in vserver / App
- precise request priority among services
- HTTP DoS protection applied to vserver / App
- Even more intelligent selection mechanism service
- Influencing decision LB for a better use of resources services
- Finally connections queued falls not if the service goes down
looks interesting, is not it? Stay tuned for the next blog retailer in the queue at the vserver.
0 Komentar