Zero Genie

The technology industry is always talking about scalability. It’s a key buzzword. “Our platform is eminently scalable and capable of adaptive accessibility across multiple change vectors,” claims just about every IT spokesperson, ever. But what does any of that really mean and why is scalability such a attractive and necessary term in the eyes of IT marketing folk?

Firstly, we can probably assume that scalability in the IT platform and application sense refers to scaling upwards, to make a piece of technology bigger and more expansive. While it’s true that the ‘flexibility-factor’ in cloud computing stems from the cloud model’s ability to allow organizations to both increase and reduce (downwards) their amount of technology consumption based upon cyclically changing needs, generally when we talk scale, we mean the drive to get bigger.

Andreas Grabner is DevOps & autonomous cloud activist at software intelligence company Dynatrace. Providing some validation for why organizations need to think about how their software behaves when the business is driving towards increased scale, Grabner explains that scalability in the modern age might be a measure of several key factors that measure how a scaling application (or other type of lower level software service) behaves.

Grabner notes that scalability could be expressed as a measure of how well a piece of software handles change in expected workload behavior situations (e.g. more traffic on Monday morning when all employees log on to their systems) or unexpected scenarios (e.g. a viral news story brings people to visit a particular website), how well it shapes up in terms of resiliency (measured in service availability) and how well a piece of software works in terms of sustaining end user experience (measured in page load times and responsiveness).

Supermarket software logic

“I like to use the example of a supermarket to explain what scalability means in the most simple terms. A supermarket can handle scale (in this case, that means more customers at 6pm after work or on Saturday mornings, or whenever customer numbers spike upwards) by putting additional cashiers on the cash registers, or it can do this by installing and firing up more automatic self-service tills. Essentially, the supermarket is able to adjust its operational approach to handle more throughput from the same core estate of business (assuming it has also provisioned for product stock levels) ensuring that customers always experience the same shopping experience (the time it takes from entering to leaving the shop) regardless of whether it is super busy, or very quiet. The parallel in software scaling comes down to whether an application or service can handle increased throughput in terms of more users, more computations, more input/output while still delivering the same performance or end user experience,” said Dynatrace’s Grabner.

Grabner is vocal on this subject because his company specializes in Application Performance Management (APM) and the higher levels of software intelligence (including AI) that can help keep a ‘full stack’ of applications and services running effectively in times of normal operation and, crucially, in times of change.

The success of an IT scalability initiative can be tracked using Service Level Indicators (SLIs) and validating them against the Service Level Objectives (SLOs) ensuring that an organization’s business-driven Service Level Agreements (SLAs) are met.

Defining scalability differentiation

Jeff Morris is vice president of product and solutions marketing at open source document-oriented database company Couchbase. Looking at the planet’s global ambitions for wider connectivity and scalability, Morris says that the dream of total scalability is a ‘sexy aspiration’ for two main reasons.

“Most of us recognize data processing demands will not stop growing until every location, device, activity and application on the planet are fully profiled, instrumented, analyzed, automated and moved to the cloud and/or whatever is beyond clouds. Software developers are constantly planning their next great innovation… and they are doing so with an expectation that it will be really big and eventually all-encompassing at massive scale,” said Morris.

Couchbase’s Morris suggests that to address these aspirations, the tech industry grows by developing and marketing around a number of key ‘change vectors’ (including processing power in the form of CPU speed, cloud, memory, storage, database, user concurrency, programming logic, website uptime, network availability etc.). He says that these measures act as ‘scalability differentiators’ because they are sustainable until that dimension hits a physical or mathematical ceiling or becomes more simply ‘commoditized’ across vendors. Looking at why we push for the next light-year leap forward all the time, Morris notes that what happens next is that the competition evolves, breaking through one barrier or another and invents their next new great differentiator.

CTO at cloud-based video management company Cloudinary Tal Lev-Ami says that he and his co-founders formed his company to solve what was essentially a scalability problem.

“The three of us had been working as consultants helping companies build web applications. One problem that kept resurfacing was that no tool existed to help developers intelligently manage large numbers of digital images. Uploading, indexing and transforming images for our clients ‘at scale’ was a tedious and time-consuming manual process. Marketing and creative teams might not realize that launching a seemingly simple ‘product sale’ campaign may require an overlay that advertises that sale on thousands of online product images, creating a huge amount of work for developers,” said Lev-Ami.

AI is the answer

Cloudinary (like a lot of technology companies looking to conquer scalability issues) is using a degree of Artificial Intelligence (AI) to help its mission. In the case of digital image and video management at scale, AI can be used to intelligently reformat videos to adapt to different environments, automatically detecting the most important visual elements. AI also helps creative teams do things like auto-tag media assets to make them easier to find and reuse. All this supports scalability by relieving web developers of having to do repetitive, manual coding jobs.   

“With every software innovation vying to be the ‘next big thing’, the need to keep the door open for expansive upward growth (while still operating successfully at a lower-scale factor) is often the mark of excellence in terms of the underlying architecture upon which any single software platform is designed.”

Jason McGee’s job focuses on knowing as much as about technology scale as possible. An IBM fellow and VP and CTO for IBM Cloud Platform, McGee says that migrating applications to the cloud can deliver significant business benefits for companies of all sizes.

“But,” advises McGee. “The ability to scale applications and ensure that they are always running and available to clients [users] is a top priority – whether it’s Black Friday, a busy travel season, or in the midst of a category five Hurricane. Multiple availability zones across all major regions of the world and the ability to scale thousands of Kubernetes clusters are key to ensuring that applications are highly available and always on — no matter the situation or circumstance.”

Whatever the weather 

McGee uses an example at The Weather Company (a business that IBM acquired in 2016), which averages 30 million-page views per day. With millions of people relying on weather.com and wunderground.com as sources of weather information, the sites must deliver petabytes of data at high speed, especially during major weather events.

“When extreme weather like a hurricane hits, pages views can spike 5x to 150 million views per day, along with 25 billion on-demand forecasts… and 40 billion API requests [to connect the forecast to other app services]. The reliability and availability of weather information in these extreme situations and their ability to perform at a high level during periods of intense demand can literally be the difference between life and death,” said IBM’s McGee.

Contemporary cloud-centric technologies today all have an eye on the scalability factor. With every software innovation vying to be the ‘next big thing’, the need to keep the door open for expansive upward growth (while still operating successfully at a lower-scale factor) is often the mark of excellence in terms of the underlying architecture upon which any single software platform is designed.

 Scalability is (if you will excuse the pun) massive.