Scalability

What does scalability stand for? It assumes that an app is capable of maintaining its productivity in the conditions of increased load. In other words, if an app withstands the increased user base without any issues, it is considered to scale well.

There are two types of scaling:

1. Vertical scaling – attracting additional capabilities (in particular, hardware capabilities) to a node (or, simply, to a computer machine);
2. Horizontal scaling – attracting additional nodes to a project (for example, adding more computer machines to a system).

As one of the dominant means of programming today, Java gives IT companies an opportunity to provide scalable development services. Being a multithreaded language, Java perfectly meets the requirement of architecture scalability to ensure task parallelization.

In addition, it is possible to choose a Java scalable web framework. Spring and Spark frameworks are among them, and they are aimed at making Java scalable solutions development easy and convenient.

One more reason why so many developers around the world prefer building scalable apps with Java technology is wide availability of Java based tools required to deliver such projects. There are both free and paid variants to choose from.

Another driver of Java popularity is the community that has been forming for years around this programming language. Thanks to this, new tools are constantly appearing and, thus, more and more developers with the help of Java design scalable applications.

To ensure scalability, a project architecture is supposed to be built with this in mind.

What are the major peculiarities that a scalable Java web application architecture should possess?

• First, it needs to be designed in such a way that, with increased load, an app will still be capable of providing prompt response to a user request;
• A common practice is to divide services into partitions, or shards – this is an example of Service-oriented architecture (SOA) implementation;
• Architecture layers often contain caches which provide faster retrieval of data that has been requested already;
• Scalable Java application architecture also has indexes that allow finding specific pieces of data among the large volumes of information;
• In addition, proxies are often inserted into a project architecture to detect repeating requests from a server and to send just one request to a database;
• One more important component of Java scalable web application architecture is a load balancer that helps distribute load when the number of user requests grows dramatically;
• To ensure asynchronous work in the condition of growing load, queues are often utilized when working on designing an architecture for a Java scalable web application.

Beside architectural specifics, there are some important considerations that should be kept in mind too when building scalable Java applications.

• It is helpful to define the conditions of probable failures in order to prevent them further;
• Load testing is useful in defining a system’s capability to operate properly under this or that load level;
• It is a good practice to define the key metrics about the app performance under the increased load and to monitor these metrics regularly.

The likelihood of considerable growth in load and number of users should ideally be planned beforehand.

That is, scalability of a project can be planned on the architecture stage. This way, once a project becomes more popular, growing number of user queries won’t be a surprise for its owner.

Alternatively, it is possible to rescale the application once the need comes and the load increases. However, programmers from almost any scalable app development company would say that the first option is preferable since the second one can be too labor-intensive and, hence, expensive. Thus, if there is a possibility to take care of a growth in load beforehand, this would be much more cost-effective.

Nevertheless, it is impossible to predict everything, and sometimes a product that was thought to be scalable fails. Thus, the existing applications often need scaling. How can this actually be achieved?

One of the approaches towards increasing the scalability of Java web application that proved to be effective is dividing a project into components and scaling them separately:

• A database;
• A user interface;
• Or any other component.

Scaling a database assumes that it should be capable of performing more operations than it did before. NoSQL databases, such as MongoDB and Apache Cassandra, as a rule, are scaled well. Based on our experience, we can say that with Apache Cassandra building scalable Java applications is usually more convenient. This database is highly scalable. How many times the number of nodes increases, the capacity grows by the same amount. By capacity we mean the quantity of requests processed per time unit.

As to scalability of Java web application user interface, it is subject to scaling too when it comes to significant load growth. The growing number of users goes in parallel with the growing variety of devices and platforms used to access an application. So, if user interface fails on new platforms appeared, it is not scaled.

Here it is better to consider not only the visual component but the interaction principles too. However, it is literally impossible to cover all the possible resolutions and sizes. Therefore, a good practice is to focus on those device characteristics that the target audience has, and, after they are covered, to proceed to the secondary ones.

As our experience of writing scalable Java applications shows, flexible layout is a principle that can be of much help. This practice assumes that certain assets can have a scalable format, so that they can be adjusted to certain platforms (and their resolutions) using programming methods.

So, even if the whole system cannot be scaled for some reason, it is still possible to scale its parts to achieve the desired performance of the product.

One more method to scale a project is to write scalable algorithms. An algorithm represents a solution for a specific issue. Hence, a scalable one is the one that does not fail when the scope of this issue increases (say, more calculations should be performed).

Obviously, if we deal with cloud hosting, issues with scalability in Java web application are extremely easy to be resolved. That is, in most cases, it is enough just to switch to another tariff in order to scale up, and additional capacity will be obtained right away.

ISS Art team knows how to build Scalable Java application solutions. One of them is a web based SaaS solution to manage construction sites. Currently, this projects handles 21’755’588 weekly queries (more than 100 of them are performed simultaneously) thanks to our scalability Java services.

Another example of Java scalability solutions delivered by our team is a prognostic tool to prevent industrial equipment malfunctions. To forecast these malfunctions, the app uses specific mathematical algorithms. To ensure that the system will maintain its good productivity with high load, we needed to design the architecture accordingly. This was achieved successfully:

• today the application forecast enables up to 25,000 components;
• the server side analysis up to 750 000 parameters.

We look forward to new exciting projects, and we are glad to propose our scaling Java application services. Our guys with 12+ years of experience in Java will prepare the architecture that will be ready for growing load. Please contact us to learn more.