1. What is web performance monitoring?
Web performance monitoring is the automatic supervision of web applications to record web performance data over time and detect error states in order to alert operations staff.
The objective of web performance monitoring is threefold:
1. Alert management – Detecting error states and alerting operations staff, so that appropriate action can be taken.
2. Data collection – Recording performance data consistently over time as an input to trouble shooting and negotiations with software and hosting suppliers.
3. Analysis & improvement – Using data to drill down, identify and remove performance bottlenecks in order to improve web site performance.
2. Why should I do it?
Many of our non-customers state that web performance is not really an issue. – “Things are going smooth, so why bother?” That’s pretty much like not putting on your safety belt because you didn’t have any accident yet.
The worst we’ve heard is an information officer saying that – “We know when our site is down anyway, because journalists call us like crazy”. If you have a more active view on your web performance and company reputation, we suggest you start monitoring your web applications.
However the real benefit of monitoring comes not from just knowing when your web site is down. If you choose a web monitoring service like Apica’s you will be able pinpoint your performance problems and raise your performance levels.
So, buckle up and start monitoring today!
3. Response time vs. uptime ?
If you are only monitoring uptime you’re only seeing half the picture. Your web application can be “up” but still be “slow”. And more importantly – When is it slow, during your valuable peak hours? In that case you could be giving the majority of your visitors poor service even though you seem to be “up” all the time.
Actually, we have a good example of how you can give poor service to 20% of your visitors and still be within the standard SLA uptime. This is what we call “The SLA Paradox”.
When you monitor response time you will know how long it takes your users to request and download one of your pages or components.
4. What makes up my response time?
Your web site’s or web application’s response time is made up of four main components.
1. Internet lookup and routing time – This is the time it takes to establish the connection between your server and the user’s computer.
2. Server wait time – Once the connection is established your servers will need time to process the request and “get ready” to start serving content (time to first byte).
3. Content loading time – This is the time from the first byte of information is sent to the user’s computer until all content is received by the user.
4. Page rendering time – Finally the user’s browser version and general computer performance will affect how quickly the page is presented, or “rendered”, in the browser window.
5. What slows down performance?
There are a thousand small things that can slow down your web site speed. That’s why you should be working tightly with a professional web performance company who has accumulated years of experience.
A short list of the most common factors for good or bad web performance would be:
1. Database/backend connections – Web sites that are heavily reliant on retrieving information from databases and various backend systems typically have lower performance.
2. Caching – One way of dealing with database generated content, is to use a front end cache. It requests and stores calls to databases, so that the content is ready for serving once the user request comes. Still, caching is no guarantee for good performance since the software can be configured in many different ways, so you will need to test and analyze your performance data to make sure your caching works properly.
3. Heavy content – The most common source of slow performance is not paying attention to the file sizes of your content. Many media rich sites and applications are simply too heavy. The best way to find out if this is a problem you have, is to perform so called “full page” tests, which give you breakdown of sizes and loading times for all the components on a page.
4. External content – Embedding external content and links on your web pages has become a standard in the Web 2.0 world. But beware - When you embed code and content of others, their web site performance will also determine your performance, since it is now a part of your site. We’ve seen hundreds of examples where a page “hangs” just because it is waiting for an ad from an adserver or a videoclip to load from some external mediaserver.
6. Using monitoring to improve web performance.
Response time monitoring is the basis for performance improvement. When you have a detailed and consistent data series of your response times, you will have the starting point for nailing down your performance problems. If you only measure uptime you won’t have so many “failure points” to examine, and it’ll be harder to find the causes of bad performance.
Many site owners have “transient” performance problems, ie. problems come and go. Monitoring response times is the perfect tool in this kind of situation. Start your monitoring and performance data starts accumulating. If a problem now occurs, there is monitoring data to analyze. You have the tools to drill down to the root cause of your performance problem and rectify it
Once that problem is fixed you will have a new bottleneck that determines your maximum performance. Repeat the procedure and your performance will be on a path of continuous improvement.
7. How is response time related to max load capacity?
Your web site performance is an indicator of your max load capacity. In general – If your response times are low and stable at low traffic volumes, there is a good chance that you will be able to handle larger volumes without problems.
But if your response times are high or if they are low, but with high variations, this is a good indicator that your web application will not be able to handle a great increase in traffic.
So a first step towards finding out if you tan take increased traffic is to investigate your response time pattern.
1. What is web performance monitoring?
Web performance monitoring is the automatic supervision of web applications to record web performance data over time and detect error states in order to alert operations staff.
The objective of web performance monitoring is threefold:
1. Alert management – Detecting error states and alerting operations staff, so that appropriate action can be taken.
2. Data collection – Recording performance data consistently over time as an input to trouble shooting and negotiations with software and hosting suppliers.
3. Analysis & improvement – Using data to drill down, identify and remove performance bottlenecks in order to improve web site performance.
2. Why should I do it?
Many of our non-customers state that web performance is not really an issue. – “Things are going smooth, so why bother?” That’s pretty much like not putting on your safety belt because you didn’t have any accident yet.
The worst we’ve heard is an information officer saying that – “We know when our site is down anyway, because journalists call us like crazy”. If you have a more active view on your web performance and company reputation, we suggest you start monitoring your web applications.
However the real benefit of monitoring comes not from just knowing when your web site is down. If you choose a web monitoring service like Apica’s you will be able pinpoint your performance problems and raise your performance levels.
So, buckle up and start monitoring today!
3. Response time vs. uptime ?
If you are only monitoring uptime you’re only seeing half the picture. Your web application can be “up” but still be “slow”. And more importantly – When is it slow, during your valuable peak hours? In that case you could be giving the majority of your visitors poor service even though you seem to be “up” all the time.
Actually, we have a good example of how you can give poor service to 20% of your visitors and still be within the standard SLA uptime. This is what we call “The SLA Paradox”.
When you monitor response time you will know how long it takes your users to request and download one of your pages or components.
4. What makes up my response time?
Your web site’s or web application’s response time is made up of four main components.
1. Internet lookup and routing time – This is the time it takes to establish the connection between your server and the user’s computer.
2. Server wait time – Once the connection is established your servers will need time to process the request and “get ready” to start serving content (time to first byte).
3. Content loading time – This is the time from the first byte of information is sent to the user’s computer until all content is received by the user.
4. Page rendering time – Finally the user’s browser version and general computer performance will affect how quickly the page is presented, or “rendered”, in the browser window.
5. What slows down performance?
There are a thousand small things that can slow down your web site speed. That’s why you should be working tightly with a professional web performance company who has accumulated years of experience.
A short list of the most common factors for good or bad web performance would be:
1. Database/backend connections – Web sites that are heavily reliant on retrieving information from databases and various backend systems typically have lower performance.
2. Caching – One way of dealing with database generated content, is to use a front end cache. It requests and stores calls to databases, so that the content is ready for serving once the user request comes. Still, caching is no guarantee for good performance since the software can be configured in many different ways, so you will need to test and analyze your performance data to make sure your caching works properly.
3. Heavy content – The most common source of slow performance is not paying attention to the file sizes of your content. Many media rich sites and applications are simply too heavy. The best way to find out if this is a problem you have, is to perform so called “full page” tests, which give you breakdown of sizes and loading times for all the components on a page.
4. External content – Embedding external content and links on your web pages has become a standard in the Web 2.0 world. But beware - When you embed code and content of others, their web site performance will also determine your performance, since it is now a part of your site. We’ve seen hundreds of examples where a page “hangs” just because it is waiting for an ad from an adserver or a videoclip to load from some external mediaserver.
6. Using monitoring to improve web performance.
Response time monitoring is the basis for performance improvement. When you have a detailed and consistent data series of your response times, you will have the starting point for nailing down your performance problems. If you only measure uptime you won’t have so many “failure points” to examine, and it’ll be harder to find the causes of bad performance.
Many site owners have “transient” performance problems, ie. problems come and go. Monitoring response times is the perfect tool in this kind of situation. Start your monitoring and performance data starts accumulating. If a problem now occurs, there is monitoring data to analyze. You have the tools to drill down to the root cause of your performance problem and rectify it
Once that problem is fixed you will have a new bottleneck that determines your maximum performance. Repeat the procedure and your performance will be on a path of continuous improvement.
7. How is response time related to max load capacity?
Your web site performance is an indicator of your max load capacity. In general – If your response times are low and stable at low traffic volumes, there is a good chance that you will be able to handle larger volumes without problems.
But if your response times are high or if they are low, but with high variations, this is a good indicator that your web application will not be able to handle a great increase in traffic.
So a first step towards finding out if you tan take increased traffic is to investigate your response time pattern.