A typical Datacenter architecture will be introduced along with the various bottlenecks it experiences. We would then look at the newer Datacenter designs and how it improves their performance/throughput. Finally, we will look at the relatively new found Networking issues like TCP Incast, TCP Outcast, etc We will start by looking at the typical Core-Access-Edge architecture and we will compare it with other architectures like Leaf-Spine, Fat-Tree etc.

More than a decade old data architecture isn’t enough for today’s data-driven businesses which are heavily dependent on AI/ML/DL. As the enterprises begin to operationalize these AI/ML workflows, they would need to optimize on storage I/O performance to feed to massively parallel GPU based compute. With growing IoT footprint, data management and AI/ML based compute challenges span across from edge, to the core and to the cloud.

Mining interesting patterns through the knowledge discovery process is an area of growing interest. Incorporating knowledge and reasoning into existing network infrastructure could result in more efficient, smart and adaptive environments. Huge volumes of data collected on a large data center network presents interesting challenges for analyzing and identifying anomalies such as network misconfiguration and impending service outages.

Multiple high performance NVMe-over-Fabrics transport options are available for greenfield deployments. But fewer options exists for those who want to leverage brownfield infrastructure. Users of FC-based flash arrays, which represents the majority of networked flash storage, can exploit a concurrent SCSI / NVMe FC SAN for low risk, minimally disruptive adoption of NVMe storage. Concurrent Fibre Channel also has the added benefit of enabling some interesting dual protocol use cases.

Clustered Samba presents the illusion of a single system image SMB server using multiple cluster nodes. The SMB protocol is difficult to implement in a cluster due to its share modes and leases. Those protocol features require clusterwide knowledge of all open file handles, contrary to a clustered NFS server for example. A file with thousands of concurrent openers presents challenges to every SMB server, even more so to a clustered one.

Clustered Samba together with ctdb have been successful together for many years now. This talk will present latest developments in Samba and ctdb from the Samba perspective. It is directed towards implementors of Samba in clustered storage products, as well as developers interested in the challenges the SMB protocol carries in clustered environments.

* Performance / Scalability
The core ctdb engine is one single threaded daemon per cluster node. Samba will take over clusterwide messaging from ctdb.

Once a ceph cluster has been assembled, how do you know the bandwidth capabilities of the cluster? What if you want to compare your cluster to one of the clusters in a vendor whitepaper? Use the Ceph Benchmarking Tool (CBT). Learn about CBT so that you can measure the performance of your cluster. You can simulate Openstack, S3 /Swift via CosBench, or an application written against the librados API. The tool is config file based with parametric sweeping to test a range of values in one file, ties in with collectl and blktrace, and will re-create a ceph cluster if desired.

Few, if any, enterprise organizations will be willing to consume an upstream version of Ceph. This session will cover general guidelines for implementing Ceph for the enterprise and cover available reference architectures from SUSE.

Learning Objectives

Understanding design choices for Ceph
General sizing guidelines
Using ceph in a traditional environment
Managing ceph in the enterprise

File system development requires to consider the multiple configurations that users can end up with. A specific setup may uncover relevant problems that do not surface on others. Storage workload and use case validation are needed (but time consuming and infrastructure failure prone) tasks within the development cycle. This session focuses on a model embedded into a Continuous Integration environment to identify file system serviced protocols (SMB, NFS, and Object) functional failures through different validation iterations.

As data-intensive workloads transition to low-latency NVMe flash-based storage to meet increasing user demand, the Fibre Channel industry is combining the lossless, highly deterministic nature of Fibre Channel with NVMe. FC-NVMe targets the performance, application response time, and scalability needed for next-generation data centers while leveraging existing Fibre Channel infrastructures.