It is not a file system, it is an object store.
It has an unlimited store capacity, it is highly scalable and available, and
exists inside a region, this means that object are not moved outside of the
region unless you specify it. By default, all objects are replicated. All
object exists in three different locations. It is a WORM store (Write Once,
Read More) so it is perfect backups, log files, etc. even for static web pages
for a web application.
Interfaces with S3 could be REST or SOAP, if we
provide a key, data will be encrypted (AES-256) automatically. The name of our
bucket could have 1024 characters as maximum, and it have to be unique (for all
AWS, not just in our region or AZ).
It enables multipart upload (100 MB/piece,
10000 pieces as maximum), access control (using Amazon S3 policies, ACLs o IAM
rules) and object versioning (you have to pay for both, original object and old
version).
Reduced Redundancy Storage (RRS) is a new
option inside S3 offering. It provides 99.99% of durability and availability,
and a low level of redundancy. You pay 2/3 of S3 price.
EBS is for AWS the same that a SAN volume for a
VMWare host. It is a simple storage volume that we could join to our EC2
instance. It is POSIX-Compliant (so you can use it as boot device). It is
replicated automatically by default inside the same Availability Zone, and
snapshots are stored in S3. A new EBS can be created from the snapshot and
attach it to another EC2.
EBS volume is “portable”, you can move it from
an instance to another (detach/attach), but it can’t be shared between different
EC2 instances.
If we choice ESB-optimized option during the
provisioning, the volume will be provisioned in a storage dedicated network
zone that has a dedicated bandwidth.
You can use EBS Provisioned IOPS too.
Differences between both solutions:
EBS Standard
|
EBS PIOPS
|
IOPS:
100 IOPS steady-state, with best-effort bursts to hundreds
|
IOPS:
Within 10% of up to 4000 IOPS, 99.9% of given year, as provisioned
|
Throughput:
variable by workload, best effort to 10s of MB/s
|
Throughput:
16 KB per I/O = up to 64 MB/s, as provisioned
|
Latency:
Varies, reads typically < 20ms writes typically < 10ms
|
Latency:
low and consistent. Second/IOPS, at recommended QD
|
Capacity:
As provisioned, up to 1 TB
|
Capacity:
As provisioned, up to 1 TB
|
In EBS Standard you pay for you use, in EBS
PIOPS you pay for provisioned.
Glacier
Similar to S3 but used for archiving purposes. Its
cost the tenth than S3 service, 1 cent/GB/month. It takes 3-5 hours to recover
the data, so you should use it for data that you don’t need in your day-by-day.
It is a WORN store (Write Once, Read Never), you pay for access to the data.
AWS Import/Export
AWS Import/Export accelerates moving large
amounts of data into and out of AWS using portable storage devices for
transport. AWS transfers your data directly onto and off of storage devices
using Amazon’s high-speed internal network and bypassing the Internet. For
significant data sets, AWS Import/Export is often faster than Internet transfer
and more cost effective than upgrading your connectivity.




No comments:
Post a Comment