What is Centralized Storage?
https://learn.flackbox.com/courses/81445/lectures/5204534
- Storage Types
- DAS – Direct Attached Storage
- Computer has dedicated storage
- Can be internal or external.
- Storage does not have it’s own compute CPU, etc.
- SAN – Storage Area Network
- NAS – Network Attached Storage
- Both provide centralized storage solutions for many computers
- DAS – Direct Attached Storage
- S & N Hardware
- Very similar to a standard server
- Mobo
- CPU
- Memory
- NIC
- Disks
- Operating system
- Smaller systems
- Disks are generally enclosed in the same chassis
- Will be built with redundancy
- Large Storage systems
- Redundant PS, controllers, etc.
- Extra Large made by clustering together
The Benefits of SAN and NAS
https://learn.flackbox.com/courses/81445/lectures/5204530
Disk Utilization with DAS
- Requires massive over provisioning to plan for expected growth
- Leave room for overhead. Adding storage often requires downtime
- Expect ~30% utilization
- Massive wasted space over several servers
- Costs very high
Using Centralized Storage
- Storage systems use a centralized ‘pool’ of shared storage
- Devices and applications can be allocated storage as required and easily changed on the fly, Non-disruptively
- Centralized storage expect ~80% utilization across all servers. Huge savings.
Thin Provisioning
- Allows you to make it appear to the servers that they have more storage than you actually paid for
- Purchased: 50 disks with 200GB = 10TB Total
- Each server thinks is has 500GB
- Over all, it looks like there is 25TB of disk, but you really only have 10TB
- Use is split up, first come, first serve
- Additional space can be added as needed, transparently to the servers
Deduplication and Compression
- Deduplication detects and eliminates identical blocks
- Eliminated blocks are replaced with a pointer to a single copy of the block on the disk
- Compression detects and eliminates redundant data and white space in files.
- Huge space savings are possible, dependent on the amount of duplicated blocks and compressible files.
Benefits
- Pooled storage moves from “just in case” to “just in time” model for purchasing
- Dedupe and compression provide additional space savings
- This provides cost savings on hardware, rack space, power and cooling
- Savings are multiplied as storage costs tends to come down over time.
Performance and Capacity
- Access can be slower than DAS since network access can add latency BUT
- Data can be striped across many disks
- Storage vendors are also at the cutting edge of new storage technologies
Resiliency
- SAN and SAS storage systems are always built to have very high degrees of resiliency
- They almost always support mission critical systems
- If a single component fails, there is a redundant component to take its place.
Centralized Management
- Much easier to manage all your storage from a centralized location rather than separately across all your servers.
Diskless Servers
- San protocols allow clients to boot up from a logical disk on the remote storage
- Client servers do not need to contain any disks
- Popular with blade servers
Storage Tiering
- Storage systems can have medial with differing attributes, such as SSD and SATA (slower but higher capacity)
- Frequently accessed (hot) data can be kept on SSD and ‘cold’ data can be archived onto the SATA drives
Centralized Backups
- Traditional DAS: 50 servers = 50 backups
- Centralized has centralized backups
- Reduces backup windows
- Does not required loading and unloading of physical media.
Snapshots
- Point in time copy of the filesystem which can be used as a convenient short term backup
- Consists of pointers to the original blocks on disk rather than being a new copy of the data, so they initially take up no space and occur nearly instantaneously.
- This is NOT a separate copy! If your storage system burns down, snapshots will also be destroyed.
- If data gets corrupted or someone accidentally deletes a file, you can quickly recover from a snapshot
- Good for quick and convenient backups and restores.
- Still need real back ups.
Disaster Recovery
- You can replicate data to a DR site
- You can load balance incoming client requests for read-only data between different sites
- You cannot do this with writable data since you need to maintain one consistent copy of the data.
Virtualization Support – vMotion
- Software such as VMware and Hyper-V allow you to run multiple virutal servers on the same underlying physical hardware server
- You can have a Linux Web server, Exchange mail server and SQL database servers all running on the same physical box, transparent to each of those virtual servers.
- Killer feature of virtualization is the ability to move virtual servers between physical servers ON THE FLY while they are still running.
- Virtual servers can keep on running with no outages even if the underlying physical server fails or is taken down for maintenance.
- External storage is a requirement for this feature.
Storage Media Type
https://learn.flackbox.com/courses/81445/lectures/5204531
RAID
Comparing SAN and NAS Storage