Sei sulla pagina 1di 33

Understanding Amazon EBS Availability and Performance

Jafar Shameem
Business Development Manager

Eric Anderson
CTO and Co-Founder, CopperEgg
Agenda

• Overview of Elastic Block Store


• Some key concepts
• Performance
• Availability
Storage Options on AWS

Block Storage Object Storage


(Elastic Block Store) (S3, Glacier)
Use for:
Use for:
• Access to raw
• Pictures, videos,
unformatted block
highly durable
level storage
media storage
• Persistent Storage
• Cold storage for
long-term archive
Amazon Elastic Block Store (EBS)
Elastic Block Storage: Persistent Storage for EC2
Feature Details
High performance Mount EBS as drives and format as
file system required

Flexible size Volumes from 1GB to 1TB in size

Secure Private to your instances


Available Replicated within an Availability Zone

Backups Volumes can be snapshotted for point in


time restore

Monitoring Detailed metrics captured via Cloud


Watch
What are some of our customers doing with EBS?
Gaming/Social/
Marketing /
Enterprises Applications Mobile/Education
Analytics

Enterprise Very high


workloads are performance
Convenient,
built on block and Fast
cost-effective,
storage consistent IO sequential IO
reliable file
for NoSQL access
Oracle, SAP, server
and relational
Microsoft DBs
Key EBS concepts
• Standard and Provisioned IOPS
Volumes
• Block Size
• Queue Depth
• Snapshots
Standard and Provisioned IOPS Volume Types

Standard Volumes Provisioned IOPS Volumes

Optimized for Workloads with low or moderate IOPS Transactional workloads requiring consistent
needs and occasional bursts. IOPS.

Volume Up to 1 TB, average 100 IOPS per volume. Up to 1TB, 2,000 IOPS per volume. Consistent
Attributes Best effort performance. Can be striped IOPS. Can be striped together for larger size and
together for larger size and higher IOPS. higher IOPS.

Workloads File server, Log processing, Websites, Business applications, MongoDB, SQL server,
Analytics, Boot, etc. MySQL, Postgres, Oracle, etc.
Block Size
Queue Depth

Maintain a number of pending I/O requests to get the most out of your Provisioned
IOPS volume. The volumes must maintain an average queue length of 1 (rounded up to
the nearest whole number) for every 200 provisioned IOPS in a minute
Snapshots
• Create snapshots (backups) of any Amazon EBS volume.
• The volume need not be attached to a running instance in order to take a
snapshot.
• These snapshots can be used to create multiple new Amazon EBS volumes,
expand the size of a volume, or move volumes across Availability Zones.
• The snapshots can be shared with specific AWS accounts or made public.
• You can use this functionality to increase the size of an existing volume, rapidly
replicate development and testing environments, or use Snapshot Copy to copy
snapshots to another region for disaster recovery or regional expansion.
Performance

• Architecting for Performance


– Avoid throughput saturation
– Striping
• Achieving Consistent Performance
– Pre-warm Provisioned IOPS volumes
– Plan for snapshot
• Snapshot Performance
Architecting for Performance: Use EBS Optimized Instances
Architecting for Performance: Avoid Throughput Saturation

• Example:
– Cluster Compute instance types have 2Gb/s bandwidth to
EBS, more than 6 PIOPS volumes at 2000 IOPS each will
saturate 2 Gb/s network
– EBS Optimized M3.2Xlarge instance has 1 Gb/s
bandwidth dedicated to EBS, more than 12 PIOPS
volumes at 500 IOPS each will saturate the 1 GB/s
network
Architecting for Performance: Striping

sudo mdadm --verbose --create /dev/md0 --level=10 --chunk=256 --raid-


devices=4 /dev/sdh1 /dev/sdh2 /dev/sdh3 /dev/sdh4
Achieving Consistent Performance: Pre-warm Provisioned
IOPS volumes

• There is a 5 to 50 percent performance reduction in IOPS when you first access


the data on a Provisioned IOPS volume.
• Write to all blocks on volumes before first use
– $ dd of=/dev/md0 if=/dev/null
Achieving Consistent Performance: Plan for Snapshot

To minimize the impact of snapshots on performance of a master node:


– create snapshots from a read replica of your data
– Plan snapshots during off-peak usage
Snapshot Performance
• To improve snapshot performance
– Increase the frequency of snapshots
Benchmarking performance
• http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/EBSPerformance.html
#benchmark_piops
• fio
• Linux, Windows
• For benchmarking I/O performance. (Note that fio has a dependency on libaio-devel.)
• Oracle ORION
• Linux, Windows
• For calibrating the I/O performance of storage systems to be used with Oracle databases.
• SQLIO
• Windows
• For calibrating the I/O performance of storage systems to be used with Microsoft SQL Server.
Testing random 4K reads
• One Volume: ~200 MongoOPS with some variability, <1mb/s
EBS • Loaded instance: ~ 1000 MongoOPS with some variability
<10mb/s


+
PIOPS •
One Volume: 2000 MongoOPS with <1% variability, 3mb/s
Loaded Instance: 20000 MongoOPS with <1% variability, 60mb/s

• Hi1.4xlarge ephemeral: ~64,000 MongoOPS with low variability, ~245mb/s


SSD
Testing random 4K reads

EBS +
PIOPS +
PIOPS SSD

Stable
Stable
Availability
• RAID
• Snapshots
RAID
• RAID 10: provides increased redundancy
– Replace EBS volume without application downtime
– Increases read throughput
– However, 50% reduction of provisioned aggregate write performance
– E.g., MongoDB optimized around the benefits of RAID 10
• RAID 0:
– All EBS volumes are replicated in the same AZ
– Increased throughput
Snapshots
CopperEgg: EBS Use Case
• How CopperEgg uses EBS
• EBS vs Provisioned IOPS EBS
• EBS and RAID
• Backup/Snapshot best practices
• Filesystem selection and tuning
• Monitoring/Migrations/Planning
How CopperEgg uses EBS
• Real-time monitoring (every 5s)
– System information
– Processes
– Synthetic HTTP/TCP/etc
– Application metrics
– Tons more..
• Requirements:
– Store many terabytes of data
– Persist the data over long periods of time
– Backups (use snapshots)
– High IO: 50-60k+ ops/s per node
• SSD + Provisioned IOPS EBS
– Consistent IO behavior (non-spikey)
EBS vs Provisioned IOPS EBS
• Standard EBS
– Good for low IO volume
– Bursty workloads may be a good
fit: do the math
• Provisioned IOPS EBS
– Great for steady IO patterns that
need consistency
– Not always more expensive than
standard!
– Be sure to use the IOPS you
provision!
EBS and RAID
• Which RAID?
– Depends on your use case, but:
• We use stripes (RAID 0) for most things
– Good performance, we build our fault tolerance at a different level
• RAID 10 (stripe of mirrors)
– Good RAID0 performance, but increase in fault tolerance due to mirrors
– Twice the cost of RAID 0
• RAID 0+1 (mirror of stripes)
– Don’t do this – same performance, worse fault tolerance
• RAID 5 (stripe with parity)
– Could be dangerous: software RAID 5 can be bad if you have any write caching enabled.
– Maybe RAID 6 (dual parity) is an option..
• Block size
– Use an appropriate stripe size for best results
• We use 64kb – but you need to test various configs to get the best fit for your application
Backup/Snapshot best practices
• Snapshot regularly
– At least once per day, more if you can
– First snapshots take a while, subsequent are faster
– Schedule for when your IO load is lowest to reduce impact
• We do it at around 9pm CST
• Use consistent naming for snapshots
– {hostname}-{raid device}-{device}-{timestamp}
• Use the API for creation
– Faster kickoff, more likely to be consistent (script it!)
– ec2-create-snapshot –d “{hostname}-{raid device}-{device}-{timestamp}” vol-d726382
• Move older snapshots to S3/Glacier for long-term storage
• RAID makes this a bit more complex:
– Make sure you unmount/snapshot/remount your file system, or use fsfreeze to keep
consistent snapshots!
Choosing a good file system
• We like ext3/4, but we love XFS
– High performance, consistent
– Robust and lots of options for tweaking/adjusting as needed
• Our favorite mount options: (your mileage may vary)
– inode64, noatime, nodiratime, attr2, nobarrier, logbufs=8, logbsize=256k, osyncisdsync, nobootwait, noauto
– Yields great performance, reduces unnecessary writes, stable
• We like ZFS a lot too, but we want to see more runtime on linux first
– But FreeBSD/ZFS would be a fine choice
• However: test your workload!
– File systems behave differently under different workloads
EBS/File system performance tuning
• Tuning file systems:
– Set the scheduler to use ‘deadline’ (for each disk in RAID array/EBS):
• [as root] echo deadline > /sys/block/[disk device]/queue/scheduler
– Adjust how aggressively the cache is written to disk. Tune these back if you are
bursty in write IO:
• vm.dirty_ratio=30
• vm.dirty_background_ratio=20
• Track what you change!
– Before changing anything, monitor it
– After you make the change, monitor it
– Then: KEEP monitoring it – things can change over time in unexpected ways
Monitoring
• Observing:
– iostat –xcd –t 1
• Watch the sum of r/s and w/s – this is your IOPS metric. For PIOPS, you want it close to the provisioned
amount. We monitor this using CopperEgg custom metrics, and alert if it goes low, or high.
– grep –A 1 dirty /proc/vmstat
• If nr_dirty approaches nr_dirty_threshold, you need to tune down vm.dirty to flush writes more often.
• Reference: http://docs.neo4j.org/chunked/stable/linux-performance-guide.html
• Useful stats to capture:
– In /proc/fs/xfs/stat
• xs_trans* -> transactions
• xs_read/write* -> read/write operations stats
• xb_* -> buffer stats
• Ignore SMART - does not work for EBS
• Watch the console log
– Use the AWS API to look for warning signs of EBS issues
Migrations and Capacity Planning
• Using PIOPS?
– Plan on a data migration path if you need to increase PIOPS
• You can’t (yet) increase IOPS on the fly
• Migration steps from an EBS backed RAID:
1. Snapshot 1hr before, then again, and again – each time it takes less time
2. Stop all services
3. Unmount the filesystem
4. Stop the RAID (mdadm –stop /dev/md0)
5. Take final snapshot
6. Create new volumes based on last snapshot
7. RAID attach new volumes – mdadm should detect the array and magically make it work.
8. Mount the filesystem
9. Restart services
Resources & Questions
• Jafar Shameem | shameemj@amazon.com | @rafaj

• http://aws.amazon.com/ebs/

• Amazon Provisioned IOPS


– http://copperegg.com/amazon-provisioned-iops-ebs/
• Benchmarking EBS performance:
– http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/EBSPerformance.html
• Stratalux: Putting Amazon’s Provisioned IOPS to the test
– http://www.stratalux.com/2012/08/09/putting-amazon’s-provisioned-iops-to-the-
test/
• MongoDB on AWS:
– http://docs.mongodb.org/ecosystem/platforms/#amazon-web-services-ec2

Potrebbero piacerti anche