HGST, now a division of Western Digital, has made first tries to save data
on smaller space using Nano-Lithography. While common technologies
allow maximum 4 to 5TB per HDD, the Nanolithography technique allows
doubling the data density. That’s quite some space for today’s demands.

The experiments were successful and the technique is as safe and sound as today’s common technology. However there will be still some time needed until first consumer produchts with the new technique will be available. The main issue might be if the writing process can still keep the pace, today’s hard disks have. Namely roughly 100 MB/s. Otherwise it would take days to fill, say, 10TB of data.

It is, however thrilling to see how much data you can already squeeze onto small harddisks, which come in sizes of 2TB at a measure of only 2.5inches. The technique is extremely small, using 10nm structure size. The image to the right shows a part of the recording surface. Also looks different from the classic HDD drives using Perpendicular recording. To get you in mind how small 10nm are: This is the size of 20 atoms side-by-side (0.5nm)!

If you remember 15 years ago, the number was the same. the Unit however was GB instead of TB. Capacities of terabytes were only achievable using large hard disk stacks in SANs or tape libraries such as the Storagetek Powderhorn.

Speaking of these, the old tape libraries (Powderhorn 9310), capable of managing around 50.000 tapes, have now been replaced by a new system called Streamline-SL8500

One of these can hold 1450 tapes with a maximum capacity of 8PB (Petabytes) uncompressed.

Since the maximum configuration can consist of 32ACL’s, each sized approx. 185.600 tapes  (I guess they allow 128 units to be combined as one ACL will store up to 1.05EB), well, you get the idea…
Each tape cartridge is capable of holding 5.5TB of data. so let’s do some math:

1’450x128x5.5 = 1’020’800 TB, = 1’012 PB = 1 EB (Exabyte) roughly rounded

Umm… quite a number here: 1’096’075’653’939’200 bytes. – and that’s just one(!) ACL)

When combining 32 of these ACLs, well…

35’074’420’926’054’400 bytes =
34’252’364’185’600 MB =
33’449’574’400 GB(!) =
32’665’600 TB(!!) =
31’900 PB(!!!) =
31.15EB

Simply said: 3.5*10¹⁶ bytes

(The tech spec says, that the ACL software can manage up to 33.8EB)

So when we switch back to the new harddisks, which shall be able to hold around 10TB of data, then you’d need 338.000 harddisks to store likewise capacity!

I don’t know the exact assembly size of only one of the 32 ACLs but a complex configuration with half the capacity already measures 7 metres depth, 18 metres width! Weighing 35 tons, this thing is as heavy as a big truck. YIKES… one would need a very large datacenter to store this gigantic configuration…

Okay, enough numbergrinding for today.

As you can see, the demand for space is there and now you might no longer wonder, why Google & Co. have such big buildings for datacenters. They even have spread them over multiple buildings as there’s often not enough space for the machinery.

Just to think about: Google appears to have approximately 180EB of data saved, this would require 6 of these giant Storagetek configurations. But since all of the data must be accessible instantly, this is more likely stored on harddisk modules that cache the tapes (VSM – Virtual storage modules) which do not have these enormous capacities available.

Would be interesting to see how big the datacentre of Google would grow if they would combine their entire data collection at one place…