Cloud Performance (includes SmartOS, Linux, and LX zones)
Location: San Francisco
Instructor: Max Bruning
This course is about cloud performance from a systems perspective, focusing on the systems that make up the environment. This covers both the challenges of traditional systems performance and also the performance of virtualization and multi-tenant environments. It is intended for system support staff, system administrators, developers, and anyone who would like to learn how systems work and perform, and the additional characteristics of the cloud. This is for both users and operators of the cloud. Topics include operating system internals, performance analysis methodology, generic application performance, CPUs, memory, file systems, disks, networking, cloud computing, and benchmarking. The example operating system studied is SmartOS, however, much of the methodologies and techniques are appropriate for others as well. Numerous performance tools are covered, from top to DTrace.
The basis for this course is the Prentice Hall book: Systems Performance: Enterprise and the Cloud. Additional labs are provided, which simulate real-world cases for the students to solve, with help from the instructor. For some students this is the highlight of the course, putting the methodologies learned into practice and seeing how they work first-hand.
systems and cloud performance
performance analysis objectives
application performance techniques, including caching, buffering, and concurrency
problems of load vs architecture
metrics: latency, IOPS, utilization, saturation
statistics: average types, multimodal distributions, outliers
essential operating system and kernel internals
identify performance features of modern kernels
how performance metrics are generated by the kernel
explain static and dynamic tracing
processor and CPU hardware architecture
kernel scheduler architecture and topics, including preemption and priority inversion
memory system architecture, including virtual memory and anonymous paging
physical memory architecture, including the MMU and MPSS
VFS and file system architecture and common performance characteristics
performance attributes of storage devices, including HDDs, SSDs, controllers, and transports
explain differences between logical and physical I/O
the role of the block device interface and the lower-level I/O stack
network latency sources
TCP performance characteristics, including buffering and listen queues
the lower level network stack
software and hardware virtualization technologies, and their architecture, overheads, and performance characteristics
resource controls used by cloud computing environments
multi-tenancy effects in cloud computing
benchmarking activities, types, and common mistakes
different performance analysis and tuning methodologies
resource and workload analysis
the problem statement method
the USE method
workload characterization method
CPU profiling, including generating flame graphs
syscall analysis using buffered tracing
solve performance issues using thread state analysis
DTrace one-liners and scripts for performance analysis
tools for CPU analysis, including vmstat, mpstat, prstat, cpustat, DTrace
tools for Memory analysis, including vmstat, prstat, pmap, DTrace
advanced memory analysis techniques, including allocation and fault tracing
tools for file system analysis, including vfsstat, truss, DTrace
advanced file system analysis techniques, including latency analysis from different stack layers
various tools for disk analysis, including iostat, iosnoop, DTrace
advanced disk analysis capabilities
tools for network analysis, including netstat, nicstat, snoop, DTrace
advanced network analysis techniques, including socket and TCP event tracing
performance analysis context for software and hardware virtualization as used in cloud computing
Familiarity with using a system via SSH, with root or equivalent privileges.
Put our containers as a service solution to the test.
Triton is an end-to-end solution that makes running containers at scale simple, secure and scalable.
Thank you for helping us improve joyent.com.
Forms are blocked
It appears you have an ad or script blocker that won't allow us to load our feedback form from app-sjf.marketo.com. To submit feedback, you can either temporarily unblock that domain, or email firstname.lastname@example.org. Email may take longer to get to the relevant people.