Salesforce Makes Its Observability Platform More Scalable

Salesforce offers businesses cloud-based software so that virtually all teams—from sales and marketing to IT—can share a single view of customer data and better connect to their customers. Its observability platform provides internal Salesforce teams with visualization tools, analytics, and notifications so that they can troubleshoot, proactively analyze how services are being used, and get ahead of potential issues for Salesforce customers.

Prior to the migration, the observability platform was running in first-party data centers where it had availability issues due to capacity challenges, making it difficult to scale and onboard new services. As the company was expanding, Salesforce Monitoring Cloud needed to onboard more of its products to the observability platform to improve availability.

In early 2021, Salesforce began migrating the observability platform from on premises to AWS. To speed up the migration and offset the cost of maintaining on-premises infrastructure while migrating to the cloud, Salesforce used the AWS Migration Acceleration Program (AWS MAP), a comprehensive and proven cloud migration program for accelerating an organization’s cloud migration and modernization journey using an outcome-driven methodology. By 2022, Salesforce had migrated over 95 percent of the services on the observability platform to AWS. “The AWS MAP program and team have been instrumental in the successful migration of the observability platform from on prem to the cloud. We are on the right architecture and platform to scale horizontally, innovate, and improve the monitoring experience at Salesforce,” says Sanjeev Lakshmanan, vice president of Salesforce Observability Platform.

During the migration and modernization effort, Salesforce worked alongside AWS teams to review and determine the right architecture for the solution and which AWS services to use. Salesforce’s observability engineers received training up front to become comfortable with AWS services and then direct support throughout the migration as issues arose. “I felt confident that we would get the support we needed from the AWS team for our migration,” says Sukumar Maddineni, senior director of engineering at Salesforce.

The Salesforce observability platform, with its on-premises architecture, needed to scale 5–10 times to handle the telemetry traffic coming from Salesforce’s growth. In the first-party footprint, the system was able to handle a few hundred million metrics per minute and a few hundred terabytes of logs per day, for example. On AWS, the observability platform ingests around two billion metrics every minute and more than 1 PB of logs. The solution evaluates nearly 250,000 alerts every minute and serves over 20,000 dashboards and tens of thousands of one-time queries from engineers in near real time. The solution can now scale as needed to fully support Salesforce’s use cases.

By migrating to AWS, Salesforce Monitoring Cloud improved availability while saving time for its staff. The platform has achieved 99.99 percent availability for alerting use cases and over 99.9 percent availability for dashboarding and one-time querying. The company is working toward achieving higher availability for all observability use cases. Instead of taking a few days to a few weeks, Salesforce Monitoring Cloud can now add capacity in 1 day to support its growing needs. Using fully managed AWS services, such as Amazon Relational Database Service (Amazon RDS)—a collection of managed services that makes it simple to set up, operate, and scale databases in the cloud—means that many issues are resolved without involvement from teams at Salesforce.

Salesforce Monitoring Cloud achieved operational efficiency with its redesigned cloud architecture, which freed up 10–15 percent of time for the team to focus on innovation, leading to increased velocity for product development. Salesforce also continues to fine-tune its build, such as by optimizing Amazon Elastic Kubernetes Service (Amazon EKS), a managed Kubernetes service to run Kubernetes on AWS and on-premises data centers. “We can refocus the bandwidth that we’re getting back on innovating and continuing to evolve our architecture,” says Maddineni.

Salesforce accelerated the migration of its observability platform in under 2 years, offsetting the cost of running both an on-premises and a cloud environment. Salesforce continues to fine-tune its architecture with support from Cloud Financial Management with AWS, which provides businesses with solutions that support cost transparency, control, forecasting, and optimization. Using AWS services, Salesforce Monitoring Cloud has continued to decrease its spend per unit and anticipate ongoing cost savings.