As ad teams grow, monoliths can become a bottleneck, slowing the delivery of optimizations and new features and increasing operational burden.
Monolith services were the primary option when many of us were trying to build large-scale infrastructure a decade back. But things have evolved quickly, and this article is going to describe some of the big reasons why microservices are beneficial from a technical, operational and development perspective.
Especially in today’s ad systems, we will look into how microservices can be tremendously beneficial for agility and faster time-to-market for products. As ad teams grow, monoliths can become a bottleneck, slowing the delivery of optimizations and new features and increasing operational burden. Below are some classic reasons to evolve from monoliths to microservices to achieve larger business goals.Monoliths are great when there are only two or three features that need to be implemented, but as that feature set grows, the code gets extremely complex and convoluted to the point that it becomes a challenge to add even a few lines of code. This is extremely relevant in ad systems, where new features must be continuously added to improve revenue and cater to advertiser requirements. When each microservice solves a specific problem in the stack, development can focus on just one service, or a small set of services, allowing new features to be added more quickly and efficiently. Thus, the addition of features becomes faster and simpler.Microservices are much easier to scale. Various features pull system resources in different directions, so it is easier to optimize and scale each microservice independently. This not only leads to higher QPS but also reduces the overall machine footprint. Moreover, binaries are hosted on machines, so smaller binaries allow for more stacking and placement opportunities, something that becomes difficult as the size of a single codebase and its binary grows. Ad services have a high machine footprint in datacenters due to the huge amount of data that passes through them. This means that any optimization could save millions and even billions of dollars in operational costs. So improving the machine footprint could be business critical.It takes more time to identify issues in a monolith due to "needle in a haystack" problems. With microservices, responsibilities are more clearly defined, and the change set is much smaller, leading to faster detection. Alerting is also easier to build and reason about. Every minute of downtime results in significant revenue loss and can impact advertiser sentiment, affecting future budget allocations due to poor ROI and ad performance. Hence, debugging and resolving issues as quickly as possible is crucial to the ad business.Small changes can be moved to production faster when working with a smaller set of services. In a large monolith, many conflicting changes depend on testing to ensure accuracy and prevent unintended impacts. In microservices, a bad bundle can be reverted by rolling back changes in just one layer of services. In ads, multiple features are often rolled out simultaneously. Rolling back production services to a previous version can unintentionally affect other features. For example, reverting Feature 1 might also roll back Feature 2, which is actively used by advertisers. While this can happen in microservices too, it is far more prominent in monoliths.The speed of development increases significantly with microservices, as individual teams work on separate codebases with fewer conflicts when deploying changes to production. This is especially important as the number of teams grows, which can otherwise become a major choke point. Faster feature development drives higher revenue by enabling larger advertiser budgets. While some changes come from core ad teams, non-ad teams also contribute heavily to the same codebase. Microservices allow these changes to be split across multiple services, enabling faster development. These are some of the key reasons observed in ad systems that prevent the delivery of revenue-critical features and slow down developers. Moving from monoliths is a multiyear effort involving numerous teams. In the meantime, teams owning these monoliths struggle to manage the codebase, coordinate with other teams and maintain the overall health of the service. It is therefore clear that investing early in migrating to a microservice architecture and building the right foundations will help prevent future slowdowns.
United States Latest News, United States Headlines
Similar News:You can also read news stories similar to this one that we have collected from other news sources.
Bar fight ends in gunfire, security guard and two others injured, suspect at largeSAN ANTONIO - A bar fight escalated into gunfire early Wednesday morning, leaving three people injured and a suspect on the run.San Antonio police officers res
Read more »
Texas girl rescued after large sand hole she was digging collapses, burying herWoodlands Fire Department in Texas calls incident a ‘near tragedy.’
Read more »
Large bear decides to call Altadena house's crawlspace ‘home'An Altadena homeowner got the surprise of his life when a large bear made itself comfortable in the crawlspace of his home.
Read more »
'So expensive': Truck drivers are shunning large pickups for smaller, compact models'Even midsize trucks have gotten too tall and too big,' one insider said.
Read more »
Volunteers clear up large illegal dumping site on road in Mustang RidgeA roadway in Mustang Ridge is clear of trash and debris after volunteers and police worked to clean an illegal dumping site.According to Mustang Ridge police, v
Read more »
Shooting at family gathering in Stockton kills 4, injures 10; shooter remains at largeCity officials said early indications suggest the shooting was targeted, and that the family gathering it happened at may have been a child's birthday party. The shooter, or possibly shooters, remains at large.
Read more »