How It Works
Granulate provides a real-time continuous performance optimization solution that can effortlessly improve server throughput 5X, reduce latency by 40%, and cut cloud compute costs by up to 60% with no code changes required.
Granulate automatically learns the application’s specific resource usage patterns and data flow to identify contended resources, bottlenecks, and prioritization opportunities. Our solution then tailors OS-level scheduling and prioritization decisions to improve the infrastructure’s application-specific performance and enable significant cost reduction.
Granulate optimizes resource allocation decision-making at the kernel and runtime levels.
Kernel optimization - Granulate optimizes kernel-level resource allocation of the main contended resources including CPU scheduling, I/O scheduling, network, and memory access, and mutexes.
Runtime optimizations - Granulate runtime-level optimizations improve the scheduling and prioritization decisions of resources managed by the runtime manager and masked from the kernel.
Thread scheduling: Custom thread scheduling, prioritized according to real-time application-specific processing stages.
Lockless networking: Optimized lockless network stack, enabling high throughput and extreme parallelism.
Inter-process communication: Efficient inter-process communication through modern protocols and shared memory that results in lower overhead and increased throughput.
Connection pooling: Intelligent connection pooling that eliminates establishment overhead without any changes to your application.
Congestion control: Autonomously crafted congestion control prioritizing between connections, adapting to the changing workload and network status.
Memory arenas: Improved memory allocations and accesses based on analyzed usage patterns, tailored to your application.
Granulate can be installed in less than a minute using a single command line. Customers can start from a single VM to see the ROI and benefits. We support all popular provisioning tools and installation methods for an automatic installation on your cluster: CLI, init ccripts, integration to machine image, Chef, Ansible, Terraform, Puppet, and Kubernetes DaemonSet.
Throughout the entire process of implementing Granulate, you require 0 code changes, 0 maintenance, and 0 configurations. Our agent is dynamically loaded onto your system, and operates at the underlying OS level, requiring no administrative or technical overhead
Granulate performs its optimization at the underlying operating system level. By doing so, we are able to tackle optimization of the technology stack from an angle that was not reached by other solutions on the market. Therefore, regardless of which best practices the customer follows, and any other cost-reduction solutions that have been deployed, Granulate helps achieve a 30-60% cost reduction benefit by optimizing new spaces across the stack.
Granulate creates an average of 30-60% cost reduction per customer workload or service. This dramatic value enables our clients to utilize more of their machines while maintaining high QoS and to easily meeting SLAs.
Check out our supported use cases and find your workload on our ROI map
Start working with Granulate by deploying Granulate gAgent ​
Try our free profiling tool and get an accurate value prediction.
​
