Organizations running on-premises workloads often have to choose between fast performance or management ease. It doesn’t have to be this way; AIOps solutions can reduce the operational burden that many teams face when managing infrastructure. At Storage Field Day in January, Tintri showcased its VMstore Intelligent Infrastructure, designed to self-optimize virtual workloads by leveraging analytics and machine learning.
Achieving Fast Performance and Management Ease with AIOps
Most organizations who run on-premises workloads want fast performance, but not by sacrificing ease of management. To actualize these goals, ITOps teams often spend a lot of time optimizing workloads and managing key components like networking and storage. But there is a better way: Artificial Intelligence for Operations or AIOps. When AIOps handles these jobs, IT teams are freed up to focus on tasks that can help grow the business.
Self-Optimizing Infrastructure for Virtual Workloads
During its Storage Field Day presentation, Tintri showcased its VMstore Intelligent Infrastructure designed to self-optimize virtual workloads by leveraging analytics and machine learning. Additionally, AI is embedded into the array. Deep integration with the filesystems and hypervisor managers like vSphere or Hyper-V lets VMstore add context and insights to these workloads beyond the underlying storage. These insights and context help VMstore to optimize performance and also make management easier.
An important distinction is Tintri’s VMstore user interface is not designed for fine-tuning the array but for highlighting what self-optimization the array has already done. Tintri considers the hands-off nature of VMstore a key part of this platform’s value proposition. Often customers must make trade-offs between performance and management ease; however, Tintri’s goal is sub-millisecond latency for its workloads through auto-tuning and self-optimization.
Reducing the Operational Burden of Managing Infrastructure
By analyzing the I/O patterns of each virtual machine and virtual disk, VMstore can determine and take action on the optimizations needed to ensure optimal and expedient performance for each workload. Through its deep integration with the underlying storage, VMstore can leverage replication, instant cloning, and snapshots at the array level but with virtual machine granularity. Functionality like instant clones of the virtual disk from one volume to another virtual machine makes an IT administrator’s job easier.
Tintri VMstore node comes in a 2U form factor. A virtual appliance manages each VMstore node, and clusters are scalable to 64 nodes. Each node sends data to the Global Center that acts as a control plane and federates the VMstore nodes for consistent and streamlined management and orchestration. Global Center supports heterogeneous VMstore management with no need to run homogenous VMstore appliances and storage media.
By centralizing the management of multiple appliances, Global Center can analyze a pool of appliances to ensure the ideal optimizations while taking advantage of the additional telemetry and derived insights. Global Center stores telemetry across the nodes on a 30-day history. Tintri has found this timeframe to be the sweet spot of making accurate predictions while minimizing the amount of telemetry data stored. Additionally, Global Center runs Monte Carlo simulations to predict events that might threaten the health of VMs over the next seven days. From these predictions, prescriptive recommendations for issues like capacity, performance, and potential caching problems are made every 12 hours. These predictions are in addition to the self-optimization that VMstore appliances make automatically.
Any technology that makes an infrastructure team’s job easier without sacrificing performance, availability, or resilience adds value to the business. Managing storage and virtual workloads are important jobs; however, many tasks can be automated and optimized. This automation and optimization can free up overburdened teams and derive better business outlines. At January’s Storage Field Day event, Tintri laid out how VMstore can self-optimize and make these goals happen.
To learn more about Tintri VMstore, check out the rest of Tintri’s Storage Field Day presentation.