1. Solid-Sate Drives Ing. Stefan Verbruggen
SSD’s
Solid-Sate Drives
Ing. Stefan Verbruggen

2. SSD
device that uses integrated circuit assemblies as memory to store data persistently (retains data when power is lost).
uses electronic interfaces compatible with traditional block input/output (I/O) hard disk drives (HDDs), which permit simple replacements in common applications
new I/O interfaces, like NVM Express have been designed to address specific requirements of the SSD technology

3. SATA and “PCI express 1x” SSD

4. SSD SSDs have no moving mechanical components
typically more resistant to physical shock
run silently
have lower access time
lower latency

5. DRAM SSD and Hybrid SSD
As of 2015, most SSDs use MLC NAND-based flash memory, which is a type of non-volatile memory that retains data when power is lost.
For applications requiring fast access but not necessarily data persistence after power loss, SSDs may be constructed from random-access memory (RAM). Such devices may employ batteries as integrated power sources to retain data for a certain amount of time after external power is lost.
Hybrid drives or solid-state hybrid drives (SSHDs) combine the features of SSDs and HDDs in the same unit, containing a large hard disk drive and an SSD cache to improve performance of frequently accessed data.[

6. NAND Floating gate transistors (FGT)
Write erase 15-20Volt to clear electrons in FGT => possible damage after (consumer) (enterprise) read/write cycles.
Read
Select bitline
Activate all word-lines, except the one to read
Word-line can be far away from bitline => noise
ECC needed !!!

7. SSD: SLC, MLC, TLC SLC: Single Level Cell
Single memory cell used to store one data-bit (2 states)
Very fast
More expensive, lasts longer
erase/write (PE) cycles/cell before failure
Server / Enterprise grade SSD’s
MLC: Multi Level Cell
4 digital states in a single cell (00,01,10,11)
fast
Cheaper, wears out faster, ECC needed
erase/write cycles/cell before failure
Client-systems / consumer grade SSD’s
TLC
8 digital states in a single cell (000 – 111)
5000 erase write cycles/cell before failure
Read/erase 3-4 times slower, Write 5-6 times slower than SLC

8. Access Times compared

9. Test / vergelijking MacBook Pro Early 2011 MacBook Pro Early2013
MacBook Pro Late2016
CPU
2GHz Core i7
2,6GHz Core i5
3,3GHz Core i7
Cores 4 2
L2/core
256KB
L3 cache
6MB
3MB
4MB
RAM
8GB 1,6GHz DDR3
16GB 2,13GHz DDR3
HDD
500GB 7200rpm
250GB SSD
500GB SSD
HD-interface
SATA – 3Gbps
SATA – 6Gbps
NVMe:
PCI-Express 4x
8 GT/s
Price incl. VAT
2.250 euro
1.600 euro
2.500 euro
Write 10GB
3min30s
30 s
5,5 s
Write speed
50MB/s
436MB/s
2049MB/s
Read 10GB
23,5 s
4,5 s

10. Test / data

11. Overprovisioning and usage tip
Failures after x write cycles for MLC en TLS
Solution: Overprovisioning
add extra space to replace failed cells
8GB on some models
Never fill a SSD to nearly completion
Little free spaces gets all the writes and wears out quickly
Solution: smart controller that ”hussels” around the more static data over the entire disk to more evenly wear out the disk (wear leveling).

12. SSD structure: grid, blocks, pages
SSD storage is a grid of blocks
Blocks consist of pages
Pages are not individually erasable
Only entire blocks are erasable
Page size: 2K, 4K, 8K, 16K
pages per block
Blocksize: 256KB – 4MB

13. Reads, writes and erasures

14. Garbage Collection (GC)
Drawback:NAND flash memory cannot overwrite existing data
They must first erase old data before writing new data to the same location.
GC is the name for the process of relocating existing data to new locations and allowing the surrounding invalid data to be erased.
Flash memory is divided into blocks, which is further divided in pages.
Data can be written directly into an empty page, but only whole blocks can be erased.
Therefore, to reclaim the space taken up by invalid data, all the valid data from one block must be first copied and written into the empty pages of a new block. Only then can the invalid data in the original block be erased, making it ready for new valid data to be written (see next slide).
NAND flash organized in een grid, bestaande uit blokken
Block bestaat uit individuele rows (pages) - Page size: 2K, 4K, 8K, 16K tot 256 pages per block => Block size from 256KB tot 4MB
Steeds volledige blok-erase door 20volt spanning die omliggende cellen stoort

15. OS AWARENESS VS DRIVE AWARENESS
In an HDD system, the Operating System (OS) can simply request that new data be written to the same location where the older, now invalid data, is stored, and the HDD will directly overwrite the old data.
In an SSD, however, the valid pages in the block must first be relocated to another block. Then the entire original block can be erased and is again ready to be used. The SSD cannot directly overwrite existing data as stated earlier (previous slides).

16. OS AWARENESS VS DRIVE AWARENESS
The OS understands the files, their structure, and the logical locations where they are stored, but does not understand the physical storage structure of the storage device.
In any storage system, the storage device doesn’t know the file structure it simply knows that there are bytes of data written in specific sectors. The storage system, whether SSD or HDD, returns the data from physical locations when the OS asks for data in the corresponding logical locations.

17. OS AWARENESS VS DRIVE AWARENESS
When the OS deletes a file, it simply marks the space used for that data as free in its logical data table.
With HDDs, the OS does not need to tell the storage device anything about the deletion because it would simply write something new into that same physical location in the future.
In the case of an SSD, it only becomes aware that the data is deleted (or invalid) when the OS tries to write to that location again. At that time the SSD marks the old data as invalid and it writes the new data to a new physical location. It may also perform GC at that same time, but that varies between SSD architectures and other conditions at that moment.

18. TRIM command
In newer operating systems, e.g., Win7, Windows Server 2008 R2, Linux , FreeBSD 8.2, Open Solaris, Mac OS X Lion, the TRIM command enables the OS to notify the SSD that old data is no longer valid about the time it deletes the logical block addresses from its logical table.
The advantage of the TRIM command is that it enables the SSD’s GC to skip the invalid data rather than moving it, thus saving time not rewriting the invalid data. This results in a reduction of the number of erase cycles on the flash memory and enables higher performance during writes.
The SSD doesn’t need to immediately delete or garbage collect these locations it just marks them as no longer valid.

19. GC without the TRIM command

20. GC with the TRIM command

21. Increase capacity further: 3D tricks
60nm > 30 nm > 15nm
?? TB

22. MLC: life extending techniques => eMLC
All implemented in the device controller
Wear leveling
moves write cycles around the chip so that cells wear evenly
Occurs during Garbage Collection
On device deduplication
reduces the volumes of data written and so lowers wear
Redundancy (overprovisioning)
reserves a portion of the device's capacity to replace cells as they fail
Write optimisation
stores data writes temporary, so they can be made in large chunks to reduce the number of write operations
=> Enterprise MLC

23. Minimize write-erase to SDD’s: tips for windoze and or OSX
Disable search and indexing (disk is fast enough)
Disable defragmenation (disk is fast enough and SSD’salways fragment through wear-leveling)
Disable paging (swap) on SSD, when enough memory is available
Disable hibernation / sleep (disk / boot is fast enough)
Disable prefetch / superfetch (XP and Vista)
Make sure TRIM is enabled !!!

24. NVMe
Non-Volatile Memory Express
Aka NVM Express or NVMe

25. NVMe Based on PCI Express (PCIe) Scalable, high bandwidth
1GB/s => 8GB/s and up

26. SSD interface evolution

27. Why PCIe for SSD’s ?

28. Tomorrow = Today

29. sources
buzz/garbage-collection-and-trim-in-ssds-explained-an- ssd-primer/2/
windows7-3.html