As digital transformation pushes companies further and further into complexity, they’ll need solutions to maintain and monitor these systems in spite of it. A service mesh offers a potential solution to a very challenging problem. However, according to the latest survey from the Cloud Native Computing Foundation (CNCF), it may not offer a final fix just yet. Let’s take a look at what this configuration offers, what the sacrifices are to get those features, and what type of companies could benefit most from the trade-off.
What does a service mesh offer?
Companies that employ a multitude of microservices need an infrastructure layer to enable service-to-service communication and to create an observability layer. It can provide service discovery capability as well as load balancing traffic.
The idea is that a service mesh provides a connection without microservices having to rewrite anything. A sidecar design pattern enables services to communicate by using a proxy. In addition, most configurations also use a control pane configuration that manages these communications.
It offers four basic features to companies—connection, security, control, and observability. Unfortunately, getting the most out of a service mesh still requires a great deal of expertise and time. Right now, it’s something most companies are interested in but may not be able to deploy just yet.
The push to adopt a service mesh depends on a few factors:
- How many microservices does a company employ?
- Are all the services written in the same language?
- How does the organization release/upgrade services?
- What type of architecture does the organization have?
Companies with a partial or total monolithic architecture and all-in-one upgrades after a manual integration process may not need to push towards a service mesh just yet. Teams that spend more time localizing a problem than fixing it or that grapple with a lack of consistent architectural pattern may find a it appropriate.
See also: Enabling Innovation with the Right Cloud Data Architecture
The buzz around a service mesh is only growing
According to the survey, a majority of the 253 survey respondents stated that they’re already in the process of developing a service mesh and are on the cusp of production. Another 19% are evaluating one. A mere nine percent stated that they weren’t considering one at all.
There’s a discrepancy in what companies search for in the service mesh solution and what’s available, but that doesn’t detract from its popularity. Companies are following along with the Kubernetes Clusters rollout to decide. A majority will run between two and ten different clusters on a service mesh in the next year.
Companies are using open source tools like Linkered to manage and deploy. As the survey notes, the scope and depth of projects depends on the features users prioritize. Security is naturally a top concern in an era of ever-increasing threats, but close behind is observability. While not always at the forefront of every conversation when it comes to innovation in tech, observability is a logical new interest as systems become more complex.
Traffic management is also of interest. Companies are concerned with ensuring that despite growing complexity, they can still process and run programs quickly and efficiently without skyrocketing costs. A service mesh hints at making this a reality. It might also help knit together endpoints and APIs to ensure no loopholes in connectivity or security as systems evolve.
Challenges to adopting the service mesh strategy come in two groups
The obstacles to a fully functional, fully realized service mesh aren’t simply technological. There are also factors in the technology spectrum that could keep companies spinning for a while. Let’s explore what’s happening.
Nontechnical obstacles can prevent adoption
Nearly half of respondents reported that a lack of talent or expertise within their organization was a significant contributing factor holding them back from full adoption. Like many digital transformation obstacles, people and talent challenges might derail efforts for a long while.
Architectural and technical complexity was the second most cited obstacle. Companies may be interested in service mesh adoption, but the field is still new enough that businesses may not have the existing infrastructure in place to support adoption quite yet. Combined with a lack of blueprints or guidance on this new concept, companies may not have a pathway to adoption despite their initial interest.
Technical obstacles are also present
Those obstacles don’t mean that technical challenges aren’t also in the way of full adoption. Integration with current tools and services is a front-running concern. However, the rest of the chart is spread relatively equally among the remaining obstacles, which included
- Reliability and consistency
- Monitoring and tracing
- Policy management
25% of respondents also noted that technical obstacles fell into a different category than the choices on the list.
Maturity in the market is still growing–and necessary
Adopting a service mesh isn’t a sure thing just yet. Companies must balance the need to manage a multitude of microservices with the continued complexity a service mesh can introduce. In some cases, setting up a service mesh could be beyond the company’s internal skills. In others, the introduction of latency could affect performance—a trade not every company is willing to make.
A service mesh is not a one size fits all approach, and there are no out-of-the-box solutions just yet. Instead, companies must decide whether a service mesh’s complexity provides a pathway for security and observability or whether it introduces unnecessary elements that can affect processing.
Ultimately, organizations may adopt service meshes more readily as the market matures and guidelines or precedents appear. Right now, the survey is clear: A service mesh is a hot topic of conversation, and one many companies are eying to solve their security and observability challenges. But it’s not a slam dunk just yet.
Elizabeth Wallace is a Nashville-based freelance writer with a soft spot for data science and AI and a background in linguistics. She spent 13 years teaching language in higher ed and now helps startups and other organizations explain – clearly – what it is they do.