Cloud migrations continue to grow, spurred on by improving technology and the aftereffects of the Covid-19 pandemic accelerating the digital economy. Gartner predicts that public cloud spending will grow from less than 17% of enterprise IT spending in 2021 to over 45% by 2026. For the IT teams performing those migrations, access to network data, specifically packet data, from business applications in the cloud is a vital proactive troubleshooting mechanism. During the application or service migration, network packet data helps IT understand how the move affects application performance and how to tune those applications for the desired user experience. After a cloud migration is complete, network packet data allows IT to continue monitoring those applications in the cloud, proactively uncover the root cause of poor application performance, and, if necessary, prove that Service Level Agreements (SLAs) with their cloud providers are not being met.
Ensuring required application performance throughout a cloud migration is crucial because customer satisfaction, competitiveness, operational efficiency, and profitability all rely on secure and responsive applications. Let’s walk through how this works.
Network packet data helps IT in troubleshoot application performance during a cloud migration in several ways:
- Baselining application performance. While planning for the application migration, packet data allows IT to baseline application performance while they are still in production on-premises. This lets them know the level of experience that their users expect so that they can maintain it throughout the migration.
- Troubleshoot during the migration. As part of the application migration, IT will need to rebuild a complex set of application dependencies and connections such as network, compute, memory, storage, services, etc. There are many opportunities for something to go wrong during this process. Packet data-based visibility allows IT to understand those dependencies, reconstruct them and test them post-migration.
- Benchmark applications in their new environment. Applications and corresponding user experiences have requirements for bandwidth, latency, and more that must be maintained in the cloud as well. Keeping an eye on packet data from those applications will show IT how those variables affect application performance and experience. For instance, is the latency on data moving from a cloud application to an on-premises database high enough that it affects user experience? Without packet data, IT can’t answer this question until users complain.
The cloud migration is the beginning of the process, not the end. Once the applications in question have been lifted and shifted to the cloud, packet data still serves a vital role in the following ways:
- Allow for ongoing monitoring. The cloud is different from on-premises data centers in many ways – applications and their dependencies interact differently, and experiences can vary. Access to network packet data in the cloud can help replicate the successful and well-tried mechanisms on-premises to tune application behaviors. It can also help provide deeper insights into how applications connect to the outside world and across different VPCs to eliminate bottlenecks.
- Determine root cause. Tracing the cause of an application performance issue in the cloud is more complicated than on-premises because IT no longer has control over the entire infrastructure – you cannot walk into an AWS, Azure, or Google data center and start rebooting servers! Application Performance Management (APM) tools alone aren’t sufficient to find and fix all types of application issues. Access to the packet data lets application and network teams collaborate and learn about deeper issues and their root causes before users complain.
- Accelerate resolution regardless of the SLA. Having access to the packet data can not only let IT track if their cloud provider is meeting their Service Level Agreement (SLA) in terms of the resources and experiences expected, but it also helps accelerate the troubleshooting and resolution process. Filing tickets and waiting for resolution with the cloud service providers can be a slow process even if IT is confident that it’s an SLA issue. IT needs to have its own visibility mechanisms to find and fix issues faster to prevent the business from suffering.
Acquiring network packet data in the cloud
Accessing high-quality packet data from the public or private cloud requires a mix of physical and/or virtual network TAPs, packet brokers, and packet capture devices, as well as traffic mirroring and load balancing categories of features. Before 2019, the big three public cloud providers were black boxes in terms of visibility, but this has changed in the last few years. Amazon Web Services (AWS) and Google Cloud now offer traffic/packet mirroring features that can mirror relevant traffic to virtual packet brokers that process and route it to performance monitoring and security tools, packet capture appliances, and any other tools that need it. In Microsoft Azure, the built-in features send packets to a virtual packet broker in “in-line” mode that copies them before sending the originals on to their destination.
Most public clouds have native monitoring services that measure different types of data, but at the moment, they only provide partial visibility and are very expensive to consume at scale. Most of those services do a reasonable job providing visibility and analytics based on log or flow data. But when it comes to the packet data, they simply provide a raw stream of packets that add little to no value. Therefore, IT needs third-party tools to consume the packet data, analyze it and create value out of it, as discussed above.
IT teams will need to decide if the benefits of a more robust monitoring infrastructure outweigh the costs and time of setting it up. But as organizations continue to move towards a cloud-first or cloud-smart model, the need to monitor business applications across the hybrid cloud will only increase. The benefits of full packet data during and after the migration process cannot be denied.
Nadeem Zahid is VP of Product Management and Marketing at cPacket Networks.