Abstract

The multitenant cloud architecture has many difficulties, one of which is the unpredictability of performance brought on by resource needs. The notion of multi-tenancy is well adopted in cloud and has taken it to a new height, but with increase in traffic, the problem of noisy neighbor originates that puts a restriction on its adoption. Therefore, a strategy needs to be devised that can address the issue of noisy neighbor without compromising upon the other SLA parameters and performance indicators. Unpredictable resource availability and usage or noise can interfere with Optimal scheduling in cloud resource scheduling, especially when resources or workloads are scheduled closely together. Current techniques for handling neighbor noise may not be sufficient and the impact of different types of noise on optimization results is unclear. The suggest system actively monitors the resource exhaustion of tenant applications and based on which can apply capabilities like assigning dedicated namespace or Utilizing a Resource Quota that can be defined on each namespace and has fixed CPU and Memory allocation will also help to reduce the noisy neighbor issue. This quota can be changed as needed to increase or decrease the resources available to each application, ensuring that no application can use up all of the cluster's compute or storage resources to the point that it interferes with neighboring applications. Making sure all applications running on a shared infrastructure have access to the resources they require at the appropriate time is the real answer to quieting loud neighbors. This is made feasible by carefully planning and sizing the infrastructure of the data center. It ought to be capable of supporting the total load at all times and have provisions for dynamic resource allocation in response to demands. Future research could focus on developing more effective noise handling techniques and exploring the impact of different types of noise on scheduling accuracy.