Daily Archives: June 7, 2015
Three seconds may not seem like a long time, but it could be the difference between making the online sale and losing a customer.
According to Radware’s STATE OF THE UNION: Ecommerce Page Speed & Web Performance – Spring 2015: “By 2010, 57% of online shoppers stated that they would abandon a web page after waiting 3 seconds for it to load. Three seconds. In case study after case study, this is the point at which most visitors will bounce if a page is not loading quickly enough. Not coincidentally, case study after case study shows that this is when business metrics – from page views to revenue – are affected by slow page rendering. Whether your goal is to convert browsers into buyers or ensure that your content is served to as many eyeballs as possible, your eye should be on this 3-second target.”
But then again, it is not only about 3 seconds, but rather about overall responsiveness, as Ron Lifton, Senior Solutions Marketing Manager, NetScout Systems, points out: “From manufacturing to highly competitive arenas such as retail banking, insurance and travel, every millisecond of responsiveness on the Web site counts.”
“Website visitors hate delay and the impact of slow response times on revenue have been well documented,” Frank Puranik, Senior Technical Specialist at iTrinegy, clarifies. “For example, Amazon calculated that a page load slowdown of just one second could cost it $1.6 billion in sales each year.”
Even for websites that are not Amazon, web response time can have a major impact on revenue. Hopefully this compilation of expert testimony will give you a little more insight into what to watch out for.
In this list, APMdigest asked industry experts – from analysts and consultants to the top vendors – to outline the most important factors that impact website response time. Each expert has given their opinion on which factor is the most significant, and the result is a well-rounded list that encompasses a wide variety of issues that can impact performance.
As often happens with these types of lists on APMdigest, many of the factors overlap and potentially fit into multiple categories. But the purpose of the list is not necessarily to relegate these issues o tidy stand-alone categories, but rather to highlight the many diverse and often interrelated factors that can impact website performance. Some of these factors are well-known issues that impact web performance, while others may open your eyes to issues you might not have thought about before.
The full list of 20 Factors That Impact Website Response Time will be posted in 4 parts over the next 4 weekdays. With factors 1–5, we start with the high level view.
Complexity is the number one factor influencing website response time. Today’s modern websites are in effect highly componentized applications built from an ever-growing mix of third-party services, cloud-based computing and self-hosted infrastructure. This rise in complexity increases potential points of failure and makes troubleshooting performance with specialized tools more challenging. This is why effective APM solutions must measure end user experience and in context of that, proactively and easily indicate what supporting infrastructure or service is inhibiting optimal quality.
VP Marketing, CA Product and Solutions Marketing, CA Technologies(link is external)
Complexity is the number one factor that impacts response time. Too often, organizations get wrapped up in adding so much functionality that performance actually suffers. Complexity can be on the client side as well as the application side. Applications can be distributed across data centers or the cloud and can utilize a variety of technologies and platforms. Code level complexity is difficult and costly to diagnose without access to tools which provide gap free data. From a customer’s perspective, it’s best to think about what a site doesn’t need – and how to simplify and streamline instead. A function that comes at the cost of performance does more harm than good.”
APM Evangelist, Dynatrace(link is external)
Fast web-response time is absolutely critical to digital business. The bourgeoning complexity of the infrastructure supporting these web applications and services has become unmanageable for many IT organizations. Without clear insight into how applications relate to infrastructure, IT lacks the visibility to assess the level of impact and to find and fix problems quickly. The result is an unpredictable and often unsatisfactory user experience. This is a pervasive problem – and one that APM solutions, which unify the perspective of both application and infrastructure are uniquely poised to solve.
President, Performance & Availability and Cloud Management/Data Center Automation, BMC Software(link is external)
Product Manager, HP Software(link is external)
Application complexity and lack of visibility create gaps where optimizations can occur. There are too many factors and often the siloed organizations and tools prevent performance tuning from occurring. Inefficiencies most often occur in the code and the database, but factors such as storage and network can be a factor.
VP of Market Development and Insights, AppDynamics(link is external)
VP, Product Marketing and Strategy for Pingdom by SolarWinds Cloud(link is external)
Although no single factor conditions the speed of website response, a key contributor is the number and latency of synchronous browser requests.
Senior Consultant, Intechnica(link is external)
The root cause behind performance issues in a web services delivery environment can be very complex and involve the network, transport, servers, service enablers (like DNS), n-tier applications, and QoS.
Senior Solutions Marketing Manager, NetScout Systems(link is external)
The top factor impacting website response time is application/infrastructure/endpoint interdependencies. Shifting dynamics across these interdependencies can cause latencies, outages, security breaches and wreak havoc on end user experience.
VP of Research, Enterprise Management Associates (EMA)(link is external)
3. CONFIGURATION AND COMMUNICATION OF COMPONENTS
Today’s website infrastructure consists of a lot of components. Some of these components aren’t even located in the same country. The installation and configuration of these components is the biggest factor of slow website response times. The response time of the website is as good as the weakest link and in my experience this usually resides in the communication and configuration of components. Gaining visibility in the configuration and the communication is crucial in detecting the root cause for these slow response times.
Online Performance Consultant and Founder of Blue Factory Internet(link is external)
The top factor that impacts website response time is latency. While mean time between failure (MTBF) and mean time to repair (MTTR) are critical metrics for front-end application performance, time to first byte (TTFB) is speed metric that drives satisfactory user experience and search rankings. TTFB is the time it takes for your browser to receive the first byte of response from a web server. A platform approach that unifies monitoring of servers and back-end infrastructure and front-end API and application performance is the key to ensuring speed and responsiveness that meet user expectations.
Technology Analyst and Founder of Tech-Tonics Advisors(link is external)
We find that latency has one of the biggest negative impacts on website response time — in other words, the distance from the website origin server to the user who is accessing the website. Organizations can get around this by either building data centers (or locating and managing servers) in many locations throughout the world or partnering with a content delivery network (CDN) that has already built a high speed network with points of presence in all of the locations the organization needs to reach customers and employees.
Director of Product Management, CDNetworks(link is external)
5. DEMAND PEAKS
Scaling is an critical factor that impacts website response time. When problems rear their ugly head it’s typically during peak times. Think Black Friday or Cyber Monday. These may be extreme examples but they illustrate a very good point. Infrastructure must be to be scaled to handle peak rates rather than average rates. Peaks in demand may only last for a short time, sometimes only milliseconds but they have a much longer lasting effect, impacting not only the web server and supporting systems but more importantly user experience. To scale the infrastructure accordingly, real-time instrumentation with sub-second granularity is key to understanding these transient peaks and the behavior of each component during these times.
Director of Technical Product Marketing, Corvil
6. WEB PAGE SIZE
In my experience, the single most consistent factor that contributes to slower load times is page size. All other things being equal, a fatter page is a slower page. In one study I conducted of 60 popular sites that used responsive design, only 20% rendered acceptably quickly — and these were also the only sites that were less than 1 MB in size. According to the HTTP Archive, the average web page today is more than 2 MB in size. This is something site owners should be aware of, and concerned about.
Senior Researcher & Evangelist, SOASTA(link is external)
7. RESPONSIVE WEB DESIGN
Mobilegeddon already is upon us where Google gives higher priority for websites that are mobile friendly. Hence, the new race is to build websites that use responsive web design techniques. This could create performance problems if resources, such as images, are not properly managed. For example, including all variations of images (for tablet, desktop, mobile, retina display) in a CSS and including this CSS incorrectly in the webpage can end up loading all variations of the image in the background for every request, hence slowing down overall page load time.
Director of Product Management, Site24x7(link is external)
VP of Acceleration, Radware(link is external)
9. WEB CONTENT AND CODE
As more and more code resides on the browser side as opposed to server side, one of the top factors impacting website response time is un-optimized browser side content like download of large images, no caching, too many redirects or DNS lookups. Using APM tools, you can see a very granular breakdown of end user response time from the browser and determine the bottleneck at a glance. You can see exactly how much time was spent on resolving DNS lookups, how much time was spent in downloading content, time spent in load event scripts and so forth. It is essential that developers fine tune their browser side code following best practices, cross test across browser types, devices and platforms, and keep an eye on the end user experience with predictive alerts to prevent slow downs.
Sr. Product Manager – APM, IBM(link is external)
The top factor that affects website response time is full page load time. Nothing harms the user experience more than slow loading times, especially when multimedia, Flash, or other graphics are involved. Users will leave your site if the video they want to view won’t load. Slow load times can be caused by content-heavy pages and poor code on the website side, or inadequate bandwidth on the user side. To ensure website performance, IT can use ping requests and loading time measurements (specifically, the time it takes to download source code) to track page speed. It is often simple to fix problems that cause slow load times, but it is critical to effectively measure what load times are across each browser and on mobile devices to understand when your website is performing poorly.
Channel Sales Manager, Paessler AG(link is external)
10. N+1 QUERIES
N+1 queries are a common web application anti-pattern that can cause slow website response time. It happens when a single application query runs other queries “automatically” – resulting in multiple queries running at once. For example, a user may be browsing through a photo gallery and while switching from one picture to another, comments get rendered along with the photos. To diagnose, N+1 issues, look into the log data – if you see one query for photos and one for “comments” generated for every photo returned by the original query, you’ve got an N+1 query on your hands.
Chief Scientist, Logentries
11. THE BACK-END SERVER
From our perspective, back-end server responsiveness is the main factor to ensure a fast website response time.
Founder and CEO, BlazeMeter(link is external)
Inadequate server resources: When the traffic to a website suddenly increases, the web server gets more requests than it can handle and fails to function efficiently. As a result, the website takes up additional server resources such as memory, disk space, bandwidth and processes, leading to a bottleneck on the server’s resources. If IT Ops don’t pay attention to the growing resource demands and plan capacity accordingly, this can lead to severe resource shortages resulting in poor-responding websites.
Sr. Market Analyst, ManageEngine(link is external)
Website response times are often tied to server performance, and when a site is slow to load, the underlying cause can be a server delay driven by network bandwidth issues such as BYOD. Because speed is a critical factor in keeping people productive with online apps, or engaged with your website, it’s imperative that organizations are proactive in monitoring servers, networks and applications to identify problems before your online end users are impacted.
VP of Product Management, Ipswitch(link is external)
While there are plethora of factors that can affect website performance, the top factor in my experience is the performance of its back-end calls. It does not matter if it is a RDBMS, an ERP system, a User directory or simply another application. If the back-end calls are slow, user experience will tank. Maybe the website is running out of database connections; maybe it is making excessive back-end calls; or maybe the back-end server is running out of hardware resources. Whatever the cause, slow back-end calls will eventually bring the website to its knees. Be sure to armor yourself with a modern APM tool to pinpoint the root cause of these slow calls.
Application Support Expert, www.karunsubramanian.com(link is external)
12. THE DATABASE
According to many reviews written by the IT Central Station community, the top factor that impacts website response time is slow database queries. Some of our reviewers recommend analyzing the queries that are executed, ensuring that they are using indexes correctly. In this way, the time spent in the database can be reduced, improving the response time.
Founder and CEO, IT Central Station(link is external)
When users experience poor website response time, IT needs to dig deeper than the network or applications. When loading, a web page might not show any errors, but users could be waiting on key dynamic content like a bank balance or an order status. This issue is frequently not the web or application server, but the database that supports them. Bad SQL code and non-optimized databases account for as many as two-thirds of all such website performance issues. Optimize SQL queries to boost website performance so people can blame the network again.
Sr. Director of Product, Idera(link is external)
One of the top factors that we’ve seen in our analysis of hundreds of applications is interactions with databases; either slow database queries or too many database queries being executed per web request. As companies analyze how their code spends it’s time, it is essential to see how often queries are called and how long they take to run.
Founder and CEO, Stackify(link is external)
13. THIRD PARTY SERVICES
Third-party integrations such as SaaS vendor plugins can have a “creeping” impact on website performance. As you accumulate more functionality via third-party vendors, you have great functionality, but there is a performance hit. Each plug-in can run multiple queries which causes longer than optimal server side processing. By understanding the performance of queries supporting your website or web application, you can take action to optimize or combine queries for improved performance.
Director, Systems Management Product Marketing, SolarWinds(link is external)
Third party vendors tend to be the biggest factor when it comes to delaying load times. The amount of requests a site has to make not only increases the time needed to load all of the elements, but also increases the chances for failure somewhere along the way. Furthermore, while reliance on third party infrastructure like DNS and CDNs can help expand the scope of your site’s user base, they also make it harder to detect isolated slowness or outages that may be localized to only a portion of your users. Without a comprehensive monitoring strategy of those third parties, you may not even know that you’re experience performance issues.
Chief Product Officer, Catchpoint(link is external)
On the client side it is counter-intuitive – when you cram sites full of “bells and whistles” to influence potential customers, problems can arise. This is because videos, pop-up chats, and pay later options are typically driven by third party applications or hosted on CDNs. Add-ons like these often create performance issues that are difficult to identify and diagnose without a robust digital performance strategy and technology that provides a comprehensive end-to-end view encapsulating 100% of every end-user experience. From a customer’s perspective, it’s best to think about what a site doesn’t need – and how to simplify and streamline instead. A function that comes at the cost of performance does more harm than good.
APM Evangelist, Dynatrace(link is external)
14. THE NETWORK
For a modern website built using today’s coding and content best practices along with sufficient host resources, acceptable user response time is only as good as the underlying network. The main variable is throughput, which is the actual amount of data that can be delivered across a network at a given time. All things being equal, the fatter the pipe the better the web performance. Throughput takes into account variables that reduce a network’s theoretical maximum transfer rate (bandwidth) including underlying physical network status, network topology, number of users, and overall network integrity. These factors can negatively impact website response time. Having performance monitoring tools that can characterize these network attributes will enable organizations to achieve optimal website performance.
Solution Manager, Network Instruments/JDSU(link is external)
Network delay for websites is the big issue. It is ultimately caused by a number of network related factors, not just bandwidth as many believe. So, website developers need to ensure they fully understand how well any site they create is going to be able to cope with the delays caused by a range of network conditions. This is going to involve conducting comprehensive pre-deployment testing in accurate replications of different network environments which is made possible, even easy, with network emulation.
Senior Technical Specialist, iTrinegy(link is external)
Over commitment of virtualized infrastructure is frequently the cause of “Performance Phantoms” that plague websites: intermittent response time spikes which have inconsistent reasons for delay. Developers chase the delay, attempting to optimize an ever-changing set of slow code identified by APM tools, yet the issue persists. In these cases the slow code is not the root cause, but rather a symptom of micro-pauses of the VM which occur when the hypervisor has a demand that exceeds available resources, so any code unlucky enough to be running during pauses gets equivalently slower. Mission critical websites running in cloud or virtualized environments should be provisioned with dedicated resources to ensure that other tenants can’t steal resources and impact website performance.
Jon C. Hodgson
Global Consulting Engineer, Riverbed(link is external)
One of the most common and yet difficult things for any admin to accomplish is to trouble shoot end user “slow-time” issues. Slow-time negatively affects enterprise performance and end user productivity ten times more often than downtime and can originate from just about anywhere within the enterprise. Constrained virtual resources can be the root cause of slow-time, or it could just be a resulting symptom. The key to resolving this issue is getting to the root cause quickly before the problem spreads to other systems bringing productivity to a standstill. The use of a 100% web based performance management solution will enable IT administrators to have universal insight into the infrastructure and enable IT to take action anytime, from anywhere on any device. The benefit is total visibility across all layers and tiers of the enterprise. Having universal insight enables and empowers IT admins to enhance IT service performance, increase operational efficiency and ensure IT effectiveness.
CEO, eG Innovations
16. IT INFRASTRUCTURE CHANGES
As new servers are powered on, database configurations changed, shared storage reconfigured, VMs reallocated, along with a whole host of other everyday infrastructure changes, little does the IT admin know how the upstream effects of these changes may be impacting web response times and the company’s bottom line. Consider this example: A storage (disk) change is made. The change slows a group of VMs. One of those VMs supports a database and therefore slows its queries. Let’s say those queries support the e-commerce application servers. As a result 60% of the users of this application experience slower responses. Or let’s say an overzealous VM admin observes that certain hosts are underutilized and adds an additional application to the underutilized servers. Now when an unexpected spike in user load occurs there will be insufficient compute resources to cover for it. Within each of those technologies, and the transitions between them, lies the potential for problems in end user transactions.
VP & GM, Dell(link is external) Performance Monitoring
17. ALTERED CODE
One of the top factors impacting website response time is altered code, which doesn’t trigger traditional monitoring alarms, as those are usually based on existing or known thresholds.
VP of Security Products, Prelert(link is external)
18. DISTRIBUTED DENIAL OF SERVICE ATTACKS (DDOS)
Quocirca researched the concerns Europe organizations have about the security and performance of their online domains and the action taken to mitigate these in 2014. The survey found, that by a small margin the biggest overall concern was denial of services attacks, which have now become so wide spread that they can effect just about any online resource. This is backed by other non-Quocirca surveys that show the number and scale of attacks has continually increased in the last few years. However, whilst it is the biggest attacks that hit the headlines, it is huge number of smaller, largely unreported attacks, that should be of most concern. These are launched as diversionary measures to mask other more targeted attacks or even as demos. DDoS was followed by user end point issues, poor network performance poor website server performance and DNS performance in that order.
Analyst and Director, Quocirca(link is external)
19. INFORMATION ARCHITECTURE
The top factor that impacts website response time is the user’s experience. And that is dependent on information architecture. Just responding fast to an http request is insufficient as that would be a technical answer to a business problem. The website exists, presumably to provide information and to answer a user’s question or questions without human intervention; thus, providing information availability 24×7. Is the information on the website structured appropriately so that the user finds the information appropriate to their need instead of waiting and then getting unhelpful information? Information architecture structures the information available for each role and may be grouped by industry. In addition good information architecture structures information in the best form to answer a questions or better yet, solve a problem. Putting the right information in front of the user without excessive navigation and false starts is the best way to improve response. The information provided should build in complexity as the user’s engagement continues. We should be timing how long it takes to get helpful information to the user and not just how long a request/response took.
VP Product Management and Marketing, Nastel Technologies(link is external)
20. THE IT TEAM
One factor that is often overlooked is finger pointing. Website response time can be impacted by a number of different factors, which can cause internal finger pointing as folks try to pinpoint the problem. Could be that the client’s network is slow; maybe there’s an issue with the WAN link out to the ISP; perhaps the firewall is slowing or denying traffic – you get the picture. Without an overarching performance monitoring platform to keep an eye on all of these disparate areas, that internal struggle can slow things down considerably.
Director of Product Marketing, SevOne(link is external)
In today’s software-defined economy where every business runs on apps, the top factor that impacts performance and response time is inattention to early warning signs like increased load time for key pages, long-running database queries, or unresolved user complaints. Often, apps provide clear indications via monitoring alerts when any of these occurs but restoring and maintaining performance first requires a culture of service quality that associates uptime with customer value. Tools and metrics are useful but only if people and process are aligned to deliver exceptional user experiences. A lack of service culture often leads to early warning signs being ignored. One way to ensure that app teams value site performance is to make it easier for them to focus on solving problems that most directly impact customer experience. Too many irrelevant alerts means valuable resources spend time figuring out what problem to solve or solving the wrong problem – either of which are inefficient and demotivating.
VP Product, Big Panda