What’s the time? Really?

av Jörgen Städje den 28 Apr 2016

Actually, very few people know. Too few. If time-critical situations can’t be accurately time-stamped, it could have a severe impact on our society. Anything from scientific projects to bus timetables and legal documents could be rendered unsafe if the time when the event occurred, the contract was signed or the credit card transaction happened, isn’t known with sufficient accuracy.

One would think that a microsecond more or less is would be of very little importance in our daily lives. Why are the metrology laboratories always adjusting their clocks a few nanoseconds, when it takes ten thousand times longer to just push a key on the keyboard or enter your PIN code?

Universal Coordinated Time

Time is nowadays produced as a global, international standard, maintained by the International Bureau of Weights and Measures (BIPM) in Paris, as an average medium of a number of atomic clocks from all over the world. In Sweden it is up to the SP Technical Research Institute of Sweden in Borås to produce Swedish Standard Time (SNT) based on French time.

How do we consume time in our computers? There are many NTP servers with associated atomic clocks in the world, and those considered reasonably accurate are allowed into the pool.ntp.org collection. That time, fine as it may be, is not traceable. It has no value in legal proceedings.

What you need is access to the Swedish time scale held by SP in Borås. With your computer connected to SP time, you will have traceable time that is legally acceptable. This is required for e.g. time-stamping legal documents and being able to prove that said document, contract or agreement, was available at the specific time, with this composition and said electronic signature.

To maintain a time service of true national value one must build a clock that can withstand high loads. It must be hard to jam or disturb. It must be verifiable. It must be able to withstand tremendous traffic volumes. It must consist of a robust mechanism, storing time in a reliable manner, more or less like a big, heavy generator generating seconds, which does not change its speed easily. That’s the NTP Project. The Swedish Post and Telecom Authority (PTS, Post- och telestyrelsen) has tasked Netnod with running the project, wherein SUNET acts as a subcontractor

This is quite a thankless task, really. It’s all about building a piece of Swedish infrastructure that no one really cares about. When it works it is completely invisible, and as such, uninteresting.

The day when the time is out of joint

The day when time stops working in our computerised society, things will be difficult to manage. Add to this the fact that no one actually knows when it stopped working. Until it is too late.

Even though the ordinary CTO dosn’t know where his servers get their time updates from, they probably get it from the NTP server at pool.ntp.org, mainly because this is the default value in most installations. Hopefully.

If the pool.ntp.org breaks, most of us wouldn’t be able to log in to our computers. Let’s suppose you make a domain login to Microsoft’s Active Directory. Most of us utilise this procedure every day, but you wouldn’t be able to login if the time difference between server and workstation would be more than five minutes. A difference of this magnitude might occur within a few days because computer real time clocks are quite bad. The same applies to the interaction between university servers and the SWAMID identity service, that will not allow login if the login certificate has the wrong time.

The ordinary user wouldn’t be able to understand what the problem is, and would stand an even smaller chance of fixing the fault.

If the world would loose its time scale, no one would notice right away, but after a few weeks we would have an international computer catastrophe.

The NTP Project

The PTS isn’t just going to sit down and watch this unfold. Instead, they tasked Netnod with building an entirely new set of clocks to be placed at Netnod interchange points in Sweden, to distribute traceable time and frequency, and operate as a time store. In its turn, Netnod tasked SUNET and others to aid in the project. It is SUNET’s task to facilitate communication between the nodes, making the solution independent of GPS. Also, there should be resources for research and development.

The present project is the second of its kind. The first project occurred ten years ago and has been running since then, but is hideously outdated. With respect to today’s threat levels, the new time stores are made with radically updated immunity to disruption.

The new clocks will be interconnected on the optical network and will be continuously compared to each other, as well as with SP standard time stored in Borås. Thus, they will form a clock ensemble, not only displaying a common time and running in synchronisation, but also being very resistant to jamming and falling out of sync.

A practical time store

tidsrack

This is what a rack of NTP servers looks like, as it is being designed and built in the SUNET office at Tulegatan in Stockholm, Sweden. This very rack will be sent off to Gothenburg to become part of the national time store. In this picture it is not yet finished, as the atomic clock is still missing.

nätverkskort

All the heavy work in the NTP server will be done by this custom designed network interface. It will receive a request packet, time stamp it, reverse the sender and receiver addresses and send it back out on the Internet again. In a hurry. That’s why everything is managed in hardware residing inside the FPGA hidden under the cooling fan at the middle of the board. The board in the image above manages a minimum of 40 gigabits per second.

Why all the frantic speed? Partly because the time stamp should not “age” too much before it is delivered back to the recipient, but also because the interface must be able to withstand extensive DDoS bombing and still be able to deliver the correct time. The card simply needs to be faster than the line it is connected to. The NTP server must not become the bottle neck, but rather the network it is connected to. The clock must be able to resist the most extreme attack scenario.

So far, Sweden hasn’t experienced foreign attacks on our time distribution, but that’s just a matter of time (!). Anyone who wishes to disrupt Swedish Standard Time will acquire the necessary resources to succeed, if it is deemed important enough.

But shouldn’t GPS be sufficient?

No. GPS is not sufficient. We put way too much trust in GPS. So much so that it can cause upheaval in the scientific community. Consider the 2010 example of the scientists at the Italian Gran Sasso laboratory, who for a moment thought that neutrino particles exceeded the speed of light. Alas, it was a mistake. By adjusting an optical fibre from a GPS receiver, the particles were reined back into Einstein’s theory of relativity.

GPS time is still not traceable to Swedish Standard Time and has no legal value. If you set your clock to GPS time you still don’t know if it is right. Setting an atomic clock is a process requiring comparing several atomic clocks to each other.

And then one day, GPS time may not exist any more, if the U.S. Department of Defence decides to block it, or because your receiver is jammed by some criminal. A jamming transmitter is only $25 on the Internet.

Should we accept this?

Criminal investigations could be hampered if CCTV cameras time-stamp their video with improper time. Credit card transactions may loose their legal value if they have bad time stamps. Currency transactions and high frequency trading could be disrupted. Trains could depart at the wring times and if mobile networks loose their synchronicity, then one day we may not be able to connect our phones to the network.

Should we have to put up with this in a technologically advanced country like Sweden?

No. That’s why we have the NTP Project.

Further reading

Only in Swedish. Please use Google Translate.

SUNET’s experimental time store in use in the computer room at Tulegatan: https://www.sunet.se/blogg/natets-centrum/

Early thoughts about the NTP Project, as they were formulated at Netnod: http://www.sweclockers.com/artikel/18531-helglasning-vad-hander-med-tiden

Early mumblings about the lack of proper real time: http://techworld.idg.se/2.2524/1.607338/brist-pa-realtid—ett-gigantiskt-hot

About NTP.org: http://ntp.org/

It is simple to jam GPS time: http://techworld.idg.se/2.2524/1.603936/varning-for-storsandare

1968-01-01

Fiberfeber startades

1 May 2015

Beställa routrar och optik

15 Sep 2015

Leverans av routrar och optik

1 Jan 2016

Byggnation av kärnnät

1 Feb 2016

Börja flytta över kunder

16 Jun 2016

Nya Sunet operativt

16 Oct 2016
2017-12-15

Fler blogginlägg av Jörgen Städje

Från oss alla, till er alla

14 Dec 2017
/ Bloggen fiberfeber

Så arbetar NOC

13 Nov 2017
/ Bloggen fiberfeber

SUNET i Hongkong

20 Sep 2017
/ Bloggen fiberfeber

SUNETs handbok i informations- och IT-säkerhet

1 Sep 2017
/ Bloggen fiberfeber

Den ökända hästen från Troja

31 Jul 2017
/ Bloggen fiberfeber

Redundans är allt

3 Jul 2017
/ Bloggen fiberfeber

SNIC-snack

2 Jun 2017
/ Bloggen fiberfeber

We have liftoff: del 5 av 2

3 Maj 2017
/ Bloggen fiberfeber

Maria Häll: We are at the Forefront!

13 Apr 2017
/ Bloggen fiberfeber

Maria Häll: Vi ligger i framkant!

10 Apr 2017
/ Bloggen fiberfeber

We have liftoff, del 4 av 2

22 Feb 2017
/ Bloggen fiberfeber

We have liftoff, del 3 av 2

30 Jan 2017
/ Bloggen fiberfeber

We have liftoff! Del 2 av 2

9 Jan 2017
/ Bloggen fiberfeber

We have liftoff! Del 1 av 2

16 Dec 2016
/ Bloggen fiberfeber

Long Read – Cleanliness is a Virtue

20 Sep 2016
/ Bloggen fiberfeber

Långläsning - tvättar bäst som tvättar först

16 Sep 2016
/ Bloggen fiberfeber

Följa fiber – från Tulegatan till Stockholms universitet.

26 Aug 2016
/ Bloggen fiberfeber

Ericsson, then swänske Lars Magnus

7 Jun 2016
/ Bloggen fiberfeber

One ring to rule them all

24 Maj 2016
/ Bloggen fiberfeber

Den tunga bakgrundstrafiken

12 Maj 2016
/ Bloggen fiberfeber

Long read: How to Design a Fibre Optic Network

5 Maj 2016
/ Bloggen fiberfeber

Welcome to the Fiber Fever Blog!

3 Maj 2016
/ Bloggen fiberfeber

Procuring an Optical Network – Smooth as Silk

2 Maj 2016
/ Blogg

The Breadth and Width of a Megabit

29 Apr 2016
/ Blogg

The Nobel Prized Piece of Glass

28 Apr 2016
/ Blogg

SUNET in i molnet (3) – molnsäkerhet

26 Apr 2016
/ Blogg

SUNET in i molnet (2) – vad är molnet egentligen?

25 Apr 2016
/ Blogg

SUNET in i molnet (1) – det här får du

25 Apr 2016
/ Blogg

Read about the brand new Sunet network.

11 Apr 2016
/ Bloggen fiberfeber

GÉANT och NORDUnet – bästa kompisar

14 Mar 2016
/ Bloggen fiberfeber

Ljuset kommer från Tyskland

3 Mar 2016
/ Bloggen fiberfeber

Thunderbirds are GO!

19 Feb 2016
/ Bloggen fiberfeber

Ett panorama av verkligheten

17 Feb 2016
/ Bloggen fiberfeber

Det allseende ögat

15 Feb 2016
/ Bloggen fiberfeber

Förstärkning på längden

15 Jan 2016
/ Bloggen fiberfeber

Dämpning och förstärkning i optisk fiber

14 Jan 2016
/ Bloggen fiberfeber

Grundläggande om L-bandet

14 Jan 2016
/ Bloggen fiberfeber

C-bandet – grundläggande om

14 Jan 2016
/ Bloggen fiberfeber

Logaritmer, min käre Watson

14 Jan 2016
/ Bloggen fiberfeber

CERN – krossen som slår sönder materiens minsta byggstenar

12 Jan 2016
/ Bloggen fiberfeber

Riksarkivets samarbete med SUNET

11 Jan 2016
/ Bloggen fiberfeber

One Ring to Rule them - Vetenskapsrådet

21 Dec 2015
/ Bloggen fiberfeber

Alla jättars jätte - Cisco

19 Dec 2015
/ Bloggen fiberfeber

En värld av siffror - belastning

19 Dec 2015
/ Bloggen fiberfeber

Ur led är inte alls tiden - atomur

19 Dec 2015
/ Bloggen fiberfeber

En djungel av kontaktdon

4 Dec 2015
/ Bloggen fiberfeber

Elektronisk enbärsdricka - Juniper

27 Nov 2015
/ Bloggen fiberfeber

Vad är Géant?

26 Nov 2015
/ Bloggen fiberfeber

Radar Love - Eiscat

25 Nov 2015
/ Bloggen fiberfeber

The Color Purple - dispersion

25 Nov 2015
/ Bloggen fiberfeber

Full Metal Packet - switchen

10 Nov 2015
/ Bloggen fiberfeber

Get your kicks on route 66 - routrar

10 Nov 2015
/ Bloggen fiberfeber

Game of Stones - kvarts

10 Nov 2015
/ Bloggen fiberfeber

The Twilight Zone - fotonen

10 Nov 2015
/ Bloggen fiberfeber

Peering – SUNETs ekonomiska ryggrad

9 Nov 2015
/ Bloggen fiberfeber

I mörkret är alla katter infraröda

4 Nov 2015
/ Bloggen fiberfeber

Fibertyperna i nätet och deras optiska felaktigheter

29 Okt 2015
/ Bloggen fiberfeber

Vad är klockan? Egentligen?

21 Okt 2015
/ Bloggen fiberfeber

Nätets centrum

20 Okt 2015
/ Bloggen fiberfeber

Den optiska transceivern

17 Okt 2015
/ Bloggen fiberfeber

Polarisation och informationsöverföring

1 Okt 2015
/ Bloggen fiberfeber

Laserns historia

30 Sep 2015
/ Bloggen fiberfeber

Koherent ljus, vad är det?

28 Sep 2015
/ Bloggen fiberfeber

När allt är klart

28 Sep 2015
/ Bloggen fiberfeber

SUNET – nu ännu bättre!

16 Sep 2015
/ Bloggen fiberfeber

Fibern fruktar fukten

11 Sep 2015
/ Bloggen fiberfeber

Att få kontakt

11 Sep 2015
/ Bloggen fiberfeber

Så tillverkas optisk fiber

31 Aug 2015
/ Bloggen fiberfeber

EMC – EMI – EMP

31 Aug 2015
/ Bloggen fiberfeber

Glasbiten som gav nobelpris

21 Aug 2015
/ Bloggen fiberfeber

Megabit på längden och tvären

21 Aug 2015
/ Bloggen fiberfeber

Långartikel: Fibern från Frostmofjället

21 Aug 2015
/ Bloggen fiberfeber

Upphandling av optiskt nät

25 Jul 2015
/ Bloggen fiberfeber

OptaSense – när fiber blir sensorer

3 Jul 2015
/ Bloggen fiberfeber

Teknisk djupdykning: Optisk magi med ramanförstärkare

2 Jul 2015
/ Bloggen fiberfeber

Teknisk utvikning: 130.000 fibrer som i en liten ask

1 Jul 2015
/ Bloggen fiberfeber

NOCen spekulerar 2: Felrapporter

27 Jun 2015
/ Bloggen fiberfeber

NOCen spekulerar 1: hög belastning

26 Jun 2015
/ Bloggen fiberfeber

Teknisk djupdykning: Optisk magi med EDFA

22 Jun 2015
/ Bloggen fiberfeber

Långartikel: Så designar man ett fiberoptiskt nät

11 Jun 2015
/ Bloggen fiberfeber

Bredare motorväg för svenska data – äntligen en offensiv satsning!

22 Maj 2015
/ Bloggen fiberfeber

Om den interaktiva tidslinjen

21 Maj 2015
/ Bloggen fiberfeber

Om den interaktiva kartan

20 Maj 2015
/ Bloggen fiberfeber

Fiberfeber: Vad som har varit och vad som komma skall

19 Maj 2015
/ Bloggen fiberfeber

Följ bygget av Sunets nät på bloggen Fiberfeber!

18 Maj 2015
/ Bloggen fiberfeber

Teknisk djupdykning: den mystiska routerkraschen

11 Jun 2006
/ Bloggen fiberfeber

2000–2013: Sunet mognar och kapaciteten ökar. Identitetsfederation skapas.

1 Jan 2000
/ Bloggen fiberfeber

1990–1999: Kapaciteten stiger, 2 – 34 – 155 Mbps

1 Jan 1990
/ Bloggen fiberfeber

1968–1989: Idéernas tidevarv. Internets vagga.

1 Jan 1968
/ Bloggen fiberfeber

Jörgen Städje

Jag heter Jörgen Städje och har skrivit om teknik och vetenskap sedan 1984. Friskt kopplat, hälften brunnet!