A number of important changes are being introduced to AzureWatch this weekend (March 16th):

• Ability to execute rules only after a sustained period of time
  • This feature allows for much better control of the scaling process.  For example, in certain situations it can be far more effective to scale when sustained load is over a certain threshold rather than try to predict what a moving average looks like.  It is important to know that if a rule is configured with a sustained time delay, it will only be executed after continuously being evaluated to TRUE for the specified period of time.
• Ability to send ON and OFF alerts (a single ON email when alert evaluates to TRUE, and a single OFF email when it evaluates to FALSE)
  • This feature reduces spam when a certain rule's condition is continuously TRUE.  It also simplifies configuration since ON/OFF alerts no longer require throttling.  Unless modified, existing Alerts will work as they currently do.
• Separation of Alerts from Management Actions (rules that notify will be separated from rules that execute scale actions, shutdowns, restarts, etc.)
  • This feature is relatively important as it may impact existing rule sets.  Going forward, rules that are Alerts will be evaluated separately from rules that are Management Actions.  When evaluating rules, all Alerts that qualify for execution will be evaluated and acted upon, not just the first one. Management Actions will continue to be evaluated until the first rule that qualifies for execution. Users who currently rely on Alerts to control execution of their Management Actions will want to revisit their scaling configurations.  We do expect percentage of such users to be either very small or non-existent.
  • After the upgrade, we plan to monitor AzureWatch's email queues and switch highly spamming Alerts to have ON/OFF logic.  Impacted customers will be notified.

While we expect minimum impact during or after the upgrade, we want to be transparent with our users: this is probably the most significant change to the Rules engine since the inception of AzureWatch. If you have any concerns, please contact Paraleap support team

The strategies discussed in this article can be applied to any cloud platform that has an ability to dynamically provision compute resources, even though I rely on examples from AzureWatch auto scaling and monitoring service for Windows Azure

The topic of auto scaling is an extremely important one when it comes to architecting cloud-based systems.  The major premise of cloud computing is its utility based approach to on-demand provisioning and de-provisioning of resources while paying only for what has been consumed.  It only makes sense to give the matter of dynamic provisioning and auto scaling a great deal of thought when designing your system to live in the cloud.  Implementing a cloud-based application without auto scaling is like installing an air-conditioner without a thermostat: one either needs to constantly monitor and manually adjust the needed temperature or pray that outside temperature never changes.

Many cloud platforms such as Amazon's EC2 or Windows Azure do not automatically adjust compute power dedicated to applications running on their platforms.  Instead, they rely upon various tools and services to provide dynamic auto scaling.  For applications running in Amazon cloud, auto-scaling is offered by Amazon itself via a service CloudWatch as well as third party vendors such as RightScale.  Windows Azure does not have its own auto scaling engine but third party vendors such as AzureWatch can provide auto scaling and monitoring.

Before deciding on when to scale up and down, it is important to understand when and why changes in demand occur.  In general, demand on your application can vary due to planned or unplanned events.  Thus it is important to initially divide your scaling strategies into these two broad categories: Predictable and Unpredictable demand.  

The goal of this article is to describe scaling strategies that gracefully handle unplanned and planned spikes in demand.  I'll use AzureWatch to demonstrate specific examples of how these strategies can be implemented in Windows Azure environment.  Important note: even though this article will mostly talk about scale up techniques, do not forget to think about matching scale down techniques. In some cases, it may help to think about building an auto scaling strategy in a way similar to building a thermostat.

Unpredictable demand

Conceptualizing use-cases of rarely occurring unplanned spikes in demand is rather straight forward.  Demand on your app may suddenly increase due to a number of various causes, such as:

• an article about your website was published on a popular website (the Slashdot effect)
• CEO of your company just ordered a number of complex reports before a big meeting with shareholders
• your marketing department just ran a successful ad campaign and forgot to tell you about the possible influx of new users
• a large overseas customer signed up overnight and started consuming a lot of resources

Whatever the case may be, having an insurance policy that deals with such unplanned spikes in demand is not just smart.  It may help save your reputation and reputation of your company.  However, gracefully handling unplanned spikes in demand can be difficult.  This is because you are reacting to events that have already happened.  There are two recommended ways of handling unplanned spikes:

Strategy 1: React to unpredictable demand

When utilization metrics are indicating high load, simply react by scaling up.  Such utilization metrics can usually include CPU utilization, amount of requests per second, number of concurrent users, amounts of bytes transferred, or amount of memory used by your application.  In AzureWatch you can configure scaling rules that aggregate such metrics over some amount of time and across all servers in the application role and then issue a scale up command when some set of averaged metrics is above a certain threshold.  In cases when multiple metrics indicate change in demand, it may also be a good idea to find a "common scaling unit", that would unify all relevant metrics together into one number.

Strategy 2: React to rate of change in unpredictable demand

Since scale-up and scale-down events take some time to execute, it may be better to interrogate the rate of increase or decrease of demand and start scaling ahead of time: when moving averages indicate acceleration or deceleration of demand.  As an example, in AzureWatch's rule-based scaling engine, such event can be represented by a rule that interrogates Average CPU utilization over a short period of time in contrast to CPU utilization over a longer period of time

(Fig: scale up when Average CPU utilization for the last 20 minutes is 20% higher than Average CPU utilization over the last hour and Average CPU utilization is already significant by being over 50%)

Also, it is important to keep in mind that scaling events with this approach may trigger at times when it is not really needed: high rate of increase will not always manifest itself in the actual demand that justifies scaling up.  However, in many instances it may be worth it to be on the safe side rather than on the cheap side.

Predictable demand

While reacting to changes in demand may be a decent insurance policy for websites with potential for unpredictable bursts in traffic, actually knowing when demand is going to be needed before it is really needed is the best way to handle auto scaling.  There are two very different ways to predict an increase or decrease in load on your application.  One way follows a pattern of demand based on historical performance and is usually schedule-based, while another is based on some sort of a "processing queue".

Strategy 3: Predictable demand based on time of day

There are frequently situations when load on the application is known ahead of time.  Perhaps it is between 7am and 7pm when a line-of-business (LOB) application is accessed by employees of a company, or perhaps it is during lunch and dinner times for an application that processes restaurant orders.  Whichever it may be, the more you know at what times during the day the demand will spike, the better off your scaling strategy will be.  AzureWatch handles this by allowing to specify scheduling aspects into execution of scaling rules.

Strategy 4: Predictable demand based on amount of work left to do

While schedule-based demand predictions are great if they exist, not all applications have consistent times of day when demand changes.  If your application utilizes some sort of a job-scheduling approach where the load on the application can be determined by the amount of jobs waiting to be processed, setting up scaling rules based on such metric may work best.  Benefits of asynchronous or batch job execution where heavy-duty processing is off-loaded to back-end servers can not only provide responsiveness and scalability to your application but also the amount of waiting-to-be-processed jobs can serve as an important leading metric in the ability to scale with better precision.  In Windows Azure, the preferred supported job-scheduling mechanism is via Queues based on Azure Storage.  AzureWatch provides an ability to create scaling rules based on the amount of messages waiting to be processed in such a queue.  For those not using Azure Queues, AzureWatch can also read custom metrics through a special XML-based interface.

Combining strategies

In the real world, implementing a combination of more than one of the above scaling strategies may be prudent.  Application administrators likely have some known patterns for their applications's behaviour that would define predictable bursting scenarios, but having an insurance policy that would handle unplanned bursts of demand may be important as well.  Understanding your demand and aligning scaling rules to work together is key to successful auto scaling implementation.

So you are deploying your application on Windows Azure cloud platform.  One of the key features of a cloud platform like Azure is the ability to consume your compute resources with a utility model.  Pay for only what is used, whether storage space, compute power, or amount of data transferred. Dynamic allocation of more space or bandwidth is built-in to Azure, but with respect to compute power, Windows Azure allows you to issue scale up or scale-down commands relatively easily.  However, deciding when to do so can be a challenging task.  This blog entry describes a service AzureWatch and how it can dynamically scale Windows Azure applications.

Part One - Introduction

At its core, AzureWatch aggregates and analyzes performance counters, queue lengths, and other metrics and matches that data against user-defined rules. When a rule produces a "hit", a scaling action or a notification occurs.  You will need an account to install and use AzureWatch.  Follow this link to fill out a simple registration form.  After registration, download link for Windows-based configuration utility will be provided.

AzureWatch currently ships in two flavors: desktop edition and server-side edition.  Server-side AzureWatch will monitor and auto-scale your Azure applications from its cloud-based servers.  Desktop edition requires a special AzureWatch Monitoring agent to be installed on your premises.  In desktop-edition the agent is responsible for gathering metrics and initiating scaling events.

Part Two - Start Control Panel

After installation is complete, start AzureWatch ControlPanel and login with your newly created account.  You will be presented with a wizard to enter your Azure connection information. 

Subscription ID can be found on your Windows Azure developer portal.  If you do not already have the X.509 certificate, AzureWatch can create one for you.  Follow the hyperlink on the wizard to get detailed instructions on creation certificates.  It is a good idea to visit AzureWatch page to understand how your certificates and storage keys are kept secure.

After entering your account SubscriptionID and specifying a valid X.509 certificate, press Connect to Azure.  You will be presented with a list of storage accounts.  Storage account that is monitored by your Diagnostics Monitor is required.

On the next wizard page you can validate default settings for such things as throttle times, notification email, etc.

After the connection wizard is completed, AzureWatch will figure out what services, deployments and roles are present.  For each role found, you will be offered a chance to create simple predefined rules.

The few sample rules offered are simple rules that rely upon basic metrics. We will come back to these rules in a short while.  For now, wizards need to be completed.

Part Three - First time in Control Panel

After wizards complete, you are presented with a dashboard screen.  It likely contains empty historical charts since no data has been collected so far.  Navigation Explorer on the left shows various parameters that can be customized, while Instructions tab on the right shows context-sensitive instructions.

It is a good idea to visit the the Rules section to see the rules that have been defined by the wizard.  Two sample rules should be present and can be edited by double-clicking on each.  Rule Edit screen is simple yet powerful.  You can specify what formula needs to be evaluated, what happens when the evaluation returns TRUE, and what time of day should evaluation be restricted to.  To make formula entry easier, a list of already defined aggregated metrics is provided.  Hovering over the formula box will display allowed operands.

One last place to visit before starting the monitoring process is the screen that contains safety limits for the number of instances that AzureWatch can scale up to or down to.  By clicking on the appropriate Role name in the navigation explorer, you will be presented with a chance to modify these boundaries.

This is it.  If you are ready, press "Publish Changes" button.  Provided your AzureWatch Monitor service is running, it will pick up these configuration settings in the next iteration of its internal loop and instruct Azure Diagnostics Manager to start capturing the metrics required for formulas to work.  Windows Azure will need a few minutes to instruct your instances to start capturing those metrics afterwards, and then a few minutes more before these metrics will be transferred to your storage.  Thus, give AzureWatch at least 5-10 minutes before expecting to see anything on the Dashboard screen.

Part Four - A few tips and tricks

Some things to keep in mind while using AzureWatch

If you just started using AzureWatch and have not accumulated enough metric data, evaluation of your rules may be suspect as your aggregations will lack sufficient data.  It maybe prudent to disable scaling inside Rules in the beginning so that your scaling actions do not trigger unexpectedly.

Metric transfer occurs only when Monitor is running.  If you stopped Monitor service for an hour and then restarted it, it does not "go back" and send the missing hour's worth of metrics to AzureWatch.

When deploying new packages into Azure, you must use do it thru the same storage account that is specified in your DiagnosticsConnectionString and in AzureWatch System Settings.  Failure to do this, will result in incomplete or incorrect metrics.

AzureWatch will always instruct your instances to capture metrics that are defined in the Raw Metrics screen.  You do not need to do anything special with existing or newly started instances.  It may be worthwhile, however, to visit the System Settings screen to further configure how metric enforcement and gathering works.

AzureWatch will send a notification when it scales your instances up or down.  In it, it will provide values for all the aggregated metrics it knows about to help you understand why the scaling event occurred.

Locally installed AzureWatch components automatically self-update whenever a new version is released.

Paraleap Technologies is proud to announce that it became a partner in Microsoft® BizSpark™ program designed to accelerate the success of emerging startups by providing key resources such as software, support and visibility.

“We are very excited to participate in the BizSpark program,” says Igor Papirov, Founder of Paraleap Technologies. "Thanks to the program we now have better access to cutting edge software and services and can concentrate on building cutting edge products targeting Microsoft Windows Azure platform"


About Paraleap Technologies

Founded in 2010, Paraleap Technologies is an emerging Chicago-based startup, focused on providing tools and services for cloud computing technologies.
AzureWatch is Paraleap’s flagship product, designed to add dynamic scalability and monitoring to applications running in Microsoft Windows Azure cloud platform.