Maintaining healthy operation of your enterprise IT infrastructure has always been central to business productivity. But, as more and more companies move their workloads to the cloud, performance monitoring has shifted towards a new set of challenges and objectives.
In traditional IT, operational metrics were essential to the smooth running of your environment—to ensure a fast and responsive service and avoid loss of productivity as a result of downtime. But the cloud has shifted the goalposts. Redundancy and scaling have reduced the risk of service outages to a near minimum. And performance issues are no longer just a matter of user experience and continuity of service. They’re now also a matter of everyday operational costs.
Turn this on its head and it’s easy to see how cloud spending behavior can actually be used as an operational metric for identifying performance issues and abnormal or malicious activity. So in this post, we highlight five common causes of sudden high cloud resource consumption that require attention at operational level.
1. Malicious Attacks
One of the most common types of cyber threat is the denial-of-service attack, which is designed to slow down or crash your server by flooding it with illegitimate requests. Typically, this can trigger unexpected auto scaling to handle the sudden workload or send your data transfer costs through the roof. Similarly, a brute force attack ramps up the load on your cloud resources, as it systematically tries different sequences of characters until it decodes an encrypted password. And if a hacker gets hold of your access keys they’ll have the potential to launch hundreds of instances and send your cloud costs skyrocketing.
The very fact malicious attacks are a potential cause of spikes underlines just how important it is to take urgent action whenever your cloud resource consumption suddenly shoots up. Because not only will it help keep the lid on your costs, but it could also prevent potential harm to your business—in the form of website downtime, loss of data or data theft.
2. Slow SQL Queries
Although SQL is a powerful database query language, it’s relatively simple to learn. But learning SQL is one thing. Optimizing it is another. Depending on the nature of the query, some database functions will be more efficient than others. And you can create two different queries that return exactly the same result but with vastly different run times.
Even a simple change to a SQL command, such as introducing a wildcard character (%) at the beginning of a LIKE pattern or retrieving more column data than needed, can suddenly bump up the cost of querying a database.
Another common pitfall to database management is indexing. Column indexes speed up queries in the same way an index makes it easier to find information in a book. An inexperienced user may not use indexes at all, so every query has to scan an entire database table. Alternatively, they may go overboard and index everything, making database updates slower and more resource intensive.
3. Inefficient Coding and Architecture
Just because a system works doesn’t necessarily mean it does it in the most efficient way. So, as with optimizing your SQL queries, you may also need to resolve inefficient coding and system architectures.
For example, your website may benefit from better handling and optimization of images, typically by using browser caching, file compression and CSS image sprites. You should also consider database caching to streamline the process of serving up dynamic web pages. And you may need to check for a new or recently updated CMS plugin, which may be responsible for slower performance and sudden increased operational costs.
And finally be aware that rehosted applications are more susceptible to operational efficiencies than those that have been rearchitected to take advantage of modern cloud features.
4. Surges in Website Traffic
Sudden bursts of website traffic may not be a direct operational issue. But how you manage them certainly is. If you use either of the two main cloud platforms, AWS and Microsoft Azure, you should enable auto scaling to handle fluctuations in traffic. Nonetheless, if your enterprise runs an e-commerce or consumer-facing website, you should be aware of seasonal purchasing patterns and forthcoming marketing campaigns, so you can accurately forecast cloud costs and avoid any unnecessary false alarms.
Not only that, but your marketing teams should also implement measures to regulate their own cloud costs. They should optimize website content and PPC ads for better targeted traffic, so they convert more visits into sales. As a result, they’ll not only increase return on their marketing investment, but they’ll also get better value for money from their cloud infrastructure expenditure.
5. Shadow IT
The advent of on-demand, pay-as-you-go cloud computing has made it easier than ever for individual business units to start their own shadow IT projects. But applications and processes, implemented without the knowledge, approval and support of the central IT department, may not be designed with operational efficiency or their deployment environment in mind.
Another issue with shadow IT is the risk to enterprise security and compliance. So if you identify a shadow IT project as the root cause of a cloud cost spike, you should view it as a potentially more serious problem.
Stay in Control
Given there are other factors, such as traffic surges to your website, cloud cost spikes are not always down to operational issues. Other causes, such as large-scale batch processing and big data analysis, are clearly normal enterprise activity. Nevertheless, you should stay on your toes and continually analyze how well operational tasks are performing. That way, you’ll avoid the more expensive alternative of scaling up your instances unnecessarily.
Also remember that some cloud costs gradually creep up without obvious signals or usage spikes. Charges for idle features, such as orphan snapshots and unused elastic IP addresses (EIPs) often go unnoticed. And running traditional antivirus software in the cloud will be more expensive than a lightweight cloud solution, which offloads most of its processing to the vendor’s own infrastructure.
The pay-as-you-go nature of on-demand computing means cloud cost spikes are inevitable. But it’s also important you quickly pinpoint the exact cause. And that’s where cloud cost monitoring and optimization tools can help—by giving you aggregated views over all your deployments and the high level of granularity you need to track those root causes down.
Start Optimizing Your Hybrid Cloud Deployment
Start your trial today to easily identify greater cloud efficiency opportunities and view your entire hybrid cloud deployment in one place. Gain full transparency, accountability, and control.