19992015 Ericsson AB. All Rights Reserved. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. Time and Time Correction in Erlang 2013-08-28 PA1 time_correction.xml
New Extended Time Functionality

As from Erlang/OTP 18 (ERTS 7.0) the time functionality has been extended. This includes a new API for time and time warp modes that change the system behavior when system time changes.

The default time warp mode has the same behavior as before, and the old API still works. Thus, you are not required to change anything unless you want to. However, you are strongly encouraged to use the new API instead of the old API based on erlang:now/0. erlang:now/0 is deprecated, as it is and will be a scalability bottleneck.

By using the new API, you automatically get scalability and performance improvements. This also enables you to use the multi-time warp mode that improves accuracy and precision of time measurements.

Terminology

To make it easier to understand this section, some terms are defined. This is a mix of our own terminology (Erlang/OS system time, Erlang/OS monotonic time, time warp) and globally accepted terminology.

Monotonically Increasing

In a monotonically increasing sequence of values, all values that have a predecessor are either larger than or equal to its predecessor.

Strictly Monotonically Increasing

In a strictly monotonically increasing sequence of values, all values that have a predecessor are larger than its predecessor.

UT1

Universal Time. UT1 is based on the rotation of the earth and conceptually means solar time at 0° longitude.

UTC

Coordinated Universal Time. UTC almost aligns with UT1. However, UTC uses the SI definition of a second, which has not exactly the same length as the second used by UT1. This means that UTC slowly drifts from UT1. To keep UTC relatively in sync with UT1, leap seconds are inserted, and potentially also deleted. That is, an UTC day can be 86400, 86401, or 86399 seconds long.

POSIX Time

Time since Epoch. Epoch is defined to be 00:00:00 UTC, 1970-01-01. A day in POSIX time is defined to be exactly 86400 seconds long. Strangely enough, Epoch is defined to be a time in UTC, and UTC has another definition of how long a day is. Quoting the Open Group "POSIX time is therefore not necessarily UTC, despite its appearance". The effect of this is that when an UTC leap second is inserted, POSIX time either stops for a second, or repeats the last second. If an UTC leap second would be deleted (which has not happened yet), POSIX time would make a one second leap forward.

Time Resolution

The shortest time interval that can be distinguished when reading time values.

Time Precision

The shortest time interval that can be distinguished repeatedly and reliably when reading time values. Precision is limited by the resolution, but resolution and precision can differ significantly.

Time Accuracy

The correctness of time values.

Time Warp

A time warp is a leap forwards or backwards in time. That is, the difference of time values taken before and after the time warp does not correspond to the actual elapsed time.

OS System Time

The operating systems view of POSIX time. To retrieve it, call os:system_time(). This may or may not be an accurate view of POSIX time. This time may typically be adjusted both backwards and forwards without limitation. That is, time warps may be observed.

To get information about the Erlang runtime system's source of OS system time, call erlang:system_info(os_system_time_source).

OS Monotonic Time

A monotonically increasing time provided by the OS. This time does not leap and has a relatively steady frequency although not completely correct. However, it is not uncommon that OS monotonic time stops if the system is suspended. This time typically increases since some unspecified point in time that is not connected to OS system time. This type of time is not necessarily provided by all OSs.

To get information about the Erlang runtime system's source of OS monotonic time, call erlang:system_info(os_monotonic_time_source).

Erlang System Time

The Erlang runtime systems view of POSIX time. To retrieve it, call erlang:system_time().

This time may or may not be an accurate view of POSIX time, and may or may not align with OS system time. The runtime system works towards aligning the two system times. Depending on the time warp mode used, this can be achieved by letting Erlang system time perform a time warp.

Erlang Monotonic Time

A monotonically increasing time provided by the Erlang runtime system. Erlang monotonic time increases since some unspecified point in time. To retrieve it, call erlang:monotonic_time().

The accuracy and precision of Erlang monotonic time heavily depends on the following:

Accuracy and precision of OS monotonic time Accuracy and precision of OS system time time warp mode used

On a system without OS monotonic time, Erlang monotonic time guarantees monotonicity, but cannot give other guarantees. The frequency adjustments made to Erlang monotonic time depend on the time warp mode used.

Internally in the runtime system, Erlang monotonic time is the "time engine" that is used for more or less everything that has anything to do with time. All timers, regardless of it is a receive ... after timer, BIF timer, or a timer in the stdlib:timer(3) module, are triggered relative Erlang monotonic time. Even Erlang system time is based on Erlang monotonic time. By adding current Erlang monotonic time with current time offset, you get current Erlang system time.

To retrieve the current time offset, call erlang:time_offset/0.

Introduction

Time is vital to an Erlang program and, more importantly, correct time is vital to an Erlang program. As Erlang is a language with soft real-time properties and we can express time in our programs, the Virtual Machine and the language must be careful about what is considered a correct time and in how time functions behave.

When Erlang was designed, it was assumed that the wall clock time in the system showed a monotonic time moving forward at exactly the same pace as the definition of time. This more or less meant that an atomic clock (or better time source) was expected to be attached to your hardware and that the hardware was then expected to be locked away from any human tinkering forever. While this can be a compelling thought, it is simply never the case.

A "normal" modern computer cannot keep time, not on itself and not unless you have a chip-level atomic clock wired to it. Time, as perceived by your computer, must normally be corrected. Hence the Network Time Protocol (NTP) protocol, together with the ntpd process, does its best to keep your computer time in sync with the correct time. Between NTP corrections, usually a less potent time-keeper than an atomic clock is used.

However, NTP is not fail-safe. The NTP server can be unavailable, ntp.conf can be wrongly configured, or your computer can sometimes be disconnected from Internet. Furthermore, you can have a user (or even system administrator) who thinks the correct way to handle Daylight Saving Time is to adjust the clock one hour two times a year (which is the incorrect way to do it). To complicate things further, this user fetched your software from Internet and has not considered what the correct time is as perceived by a computer. The user does not care about keeping the wall clock in sync with the correct time. The user expects your program to have unlimited knowledge about the time.

Most programmers also expect time to be reliable, at least until they realize that the wall clock time on their workstation is off by a minute. Then they set it to the correct time, but most probably not in a smooth way.

The number of problems that arise when you always expect the wall clock time on the system to be correct can be immense. Erlang therefore introduced the "corrected estimate of time", or the "time correction", many years ago. The time correction relies on the fact that most operating systems have some kind of monotonic clock, either a real-time extension or some built-in "tick counter" that is independent of the wall clock settings. This counter can have microsecond resolution or much less, but it has a drift that cannot be ignored.

Time Correction

If time correction is enabled, the Erlang runtime system makes use of both OS system time and OS monotonic time, to adjust the frequency of the Erlang monotonic clock. Time correction ensures that Erlang monotonic time does not warp and that the frequency is relatively accurate. The type of frequency adjustments depends on the time warp mode used. Section Time Warp Modes provides more details.

By default time correction is enabled if support for it exists on the specific platform. Support for it includes both OS monotonic time, provided by the OS, and an implementation in the Erlang runtime system using OS monotonic time. To check if your system has support for OS monotonic time, call erlang:system_info(os_monotonic_time_source). To check if time correction is enabled on your system, call erlang:system_info(time_correction).

To enable or disable time correction, pass command-line argument +c [true|false] to erl(1).

If time correction is disabled, Erlang monotonic time can warp forwards or stop, or even freeze for extended periods of time. There are then no guarantees that the frequency of the Erlang monotonic clock is accurate or stable.

You typically never want to disable time correction. Previously a performance penalty was associated with time correction, but nowadays it is usually the other way around. If time correction is disabled, you probably get bad scalability, bad performance, and bad time measurements.

Time Warp Safe Code

Time warp safe code can handle a time warp of Erlang system time.

erlang:now/0 behaves bad when Erlang system time warps. When Erlang system time does a time warp backwards, the values returned from erlang:now/0 freeze (if you disregard the microsecond increments made because of the actual call) until OS system time reaches the point of the last value returned by erlang:now/0. This freeze can continue for a long time. It can take years, decades, and even longer until the freeze stops.

All uses of erlang:now/0 are not necessarily time warp unsafe. If you do not use it to get time, it is time warp safe. However, all uses of erlang:now/0 are suboptimal from a performance and scalability perspective. So you really want to replace the use of it with other functionality. For examples of how to replace the use of erlang:now/0, see section How to Work with the New API.

Time Warp Modes

Current Erlang system time is determined by adding the current Erlang monotonic time with current time offset. The time offset is managed differently depending on which time warp mode you use.

To set the time warp mode, pass command-line argument +C [no_time_warp|single_time_warp|multi_time_warp] to erl(1).

No Time Warp Mode

The time offset is determined at runtime system start and does not change later. This is the default behavior, but not because it is the best mode (which it is not). It is default only because this is how the runtime system behaved until ERTS 7.0. Ensure that your Erlang code that can execute during a time warp is time warp safe before enabling other modes.

As the time offset is not allowed to change, time correction must adjust the frequency of the Erlang monotonic clock to align Erlang system time with OS system time smoothly. A significant downside of this approach is that we on purpose will use a faulty frequency on the Erlang monotonic clock if adjustments are needed. This error can be as large as 1%. This error will show up in all time measurements in the runtime system.

If time correction is not enabled, Erlang monotonic time freezes when OS system time leaps backwards. The freeze of monotonic time continues until OS system time catches up. The freeze can continue for a long time. When OS system time leaps forwards, Erlang monotonic time also leaps forward.

Single Time Warp Mode

This mode is more or less a backward compatibility mode as from its introduction.

On an embedded system it is not uncommon that the system has no power supply, not even a battery, when it is shut off. The system clock on such a system is typically way off when the system boots. If no time warp mode is used, and the Erlang runtime system is started before OS system time has been corrected, Erlang system time can be wrong for a long time, centuries or even longer.

If you need to use Erlang code that is not time warp safe, and you need to start the Erlang runtime system before OS system time has been corrected, you may want to use the single time warp mode.

There are limitations to when you can execute time warp unsafe code using this mode. If it is possible to use time warp safe code only, it is much better to use the multi-time warp mode instead.

Using the single time warp mode, the time offset is handled in two phases:

Preliminary Phase

This phase starts when the runtime system starts. A preliminary time offset based on current OS system time is determined. This offset is from now on to be fixed during the whole preliminary phase.

If time correction is enabled, adjustments to the Erlang monotonic clock are made to keep its frequency as correct as possible. However, no adjustments are made trying to align Erlang system time and OS system time. That is, during the preliminary phase Erlang system time and OS system time can diverge from each other, and no attempt is made to prevent this.

If time correction is disabled, changes in OS system time affects the monotonic clock the same way as when the no time warp mode is used.

Final Phase

This phase begins when the user finalizes the time offset by calling erlang:system_flag(time_offset, finalize). The finalization can only be performed once.

During finalization, the time offset is adjusted and fixed so that current Erlang system time aligns with the current OS system time. As the time offset can change during the finalization, Erlang system time can do a time warp at this point. The time offset is from now on fixed until the runtime system terminates. If time correction has been enabled, the time correction from now on also makes adjustments to align Erlang system time with OS system time. When the system is in the final phase, it behaves exactly as in no time warp mode.

In order for this to work properly, the user must ensure that the following two requirements are satisfied:

Forward Time Warp

The time warp made when finalizing the time offset can only be done forwards without encountering problems. This implies that the user must ensure that OS system time is set to a time earlier or equal to actual POSIX time before starting the Erlang runtime system.

If you are not sure that OS system time is correct, set it to a time that is guaranteed to be earlier than actual POSIX time before starting the Erlang runtime system, just to be safe.

Finalize Correct OS System Time

OS system time must be correct when the user finalizes the time offset.

If these requirements are not fulfilled, the system may behave very bad.

Assuming that these requirements are fulfilled, time correction is enabled, and OS system time is adjusted using a time adjustment protocol such as NTP, only small adjustments of Erlang monotonic time are needed to keep system times aligned after finalization. As long as the system is not suspended, the largest adjustments needed are for inserted (or deleted) leap seconds.

To use this mode, ensure that all Erlang code that will execute in both phases is time warp safe.

Code executing only in the final phase does not have to be able to cope with the time warp.

Multi-Time Warp Mode

Multi-time warp mode in combination with time correction is the preferred configuration. This as the Erlang runtime system have better performance, scale better, and behave better on almost all platforms. Also, the accuracy and precision of time measurements are better. Only Erlang runtime systems executing on ancient platforms benefit from another configuration.

The time offset can change at any time without limitations. That is, Erlang system time can perform time warps both forwards and backwards at any time. As we align Erlang system time with OS system time by changing the time offset, we can enable a time correction that tries to adjust the frequency of the Erlang monotonic clock to be as correct as possible. This makes time measurements using Erlang monotonic time more accurate and precise.

If time correction is disabled, Erlang monotonic time leaps forward if OS system time leaps forward. If OS system time leaps backwards, Erlang monotonic time stops briefly, but it does not freeze for extended periods of time. This as the time offset is changed to align Erlang system time with OS system time.

To use this mode, ensure that all Erlang code that will execute on the runtime system is time warp safe.

New Time API

The old time API is based on erlang:now/0. erlang:now/0 was intended to be used for many unrelated things. This tied these unrelated operations together and caused issues with performance, scalability, accuracy, and precision for operations that did not need to have such issues. To improve this, the new API spreads different functionality over multiple functions.

To be backward compatible, erlang:now/0 remains "as is", but you are strongly discouraged from using it. Many use cases of erlang:now/0 prevents you from using the new multi-time warp mode, which is an important part of this new time functionality improvement.

Some of the new BIFs on some systems, perhaps surprisingly, return negative integer values on a newly started runtime system. This is not a bug, but a memory use optimization.

The new API consists of the following new BIFs:

erlang:convert_time_unit/3

erlang:monotonic_time/0

erlang:monotonic_time/1

erlang:system_time/0

erlang:system_time/1

erlang:time_offset/0

erlang:time_offset/1

erlang:timestamp/0

erlang:unique_integer/0

erlang:unique_integer/1

os:system_time/0

os:system_time/1

The new API also consists of extensions of the following existing BIFs:

erlang:monitor(time_offset, clock_service)

erlang:system_flag(time_offset, finalize)

erlang:system_info(os_monotonic_time_source)

erlang:system_info(os_system_time_source)

erlang:system_info(time_offset)

erlang:system_info(time_warp_mode)

erlang:system_info(time_correction)

erlang:system_info(start_time)

erlang:system_info(end_time)

New Erlang Monotonic Time

Erlang monotonic time as such is new as from ERTS 7.0. It is introduced to detach time measurements, such as elapsed time from calendar time. In many use cases there is a need to measure elapsed time or specify a time relative to another point in time without the need to know the involved times in UTC or any other globally defined time scale. By introducing a time scale with a local definition of where it starts, time that do not concern calendar time can be managed on that time scale. Erlang monotonic time uses such a time scale with a locally defined start.

The introduction of Erlang monotonic time allows us to adjust the two Erlang times (Erlang monotonic time and Erlang system time) separately. By doing this, the accuracy of elapsed time does not have to suffer just because the system time happened to be wrong at some point in time. Separate adjustments of the two times are only performed in the time warp modes, and only fully separated in the multi-time warp mode. All other modes than the multi-time warp mode are for backward compatibility reasons. When using these modes, the accuracy of Erlang monotonic time suffer, as the adjustments of Erlang monotonic time in these modes are more or less tied to Erlang system time.

The adjustment of system time could have been made smother than using a time warp approach, but we think that would be a bad choice. As we can express and measure time that is not connected to calendar time by the use of Erlang monotonic time, it is better to expose the change in Erlang system time immediately. This as the Erlang applications executing on the system can react on the change in system time as soon as possible. This is also more or less exactly how most operating systems handle this (OS monotonic time and OS system time). By adjusting system time smoothly, we would just hide the fact that system time changed and make it harder for the Erlang applications to react to the change in a sensible way.

To be able to react to a change in Erlang system time, you must be able to detect that it happened. The change in Erlang system time occurs when the current time offset is changed. We have therefore introduced the possibility to monitor the time offset using erlang:monitor(time_offset, clock_service). A process monitoring the time offset is sent a message on the following format when the time offset is changed:

{'CHANGE', MonitorReference, time_offset, clock_service, NewTimeOffset}
Unique Values

Besides reporting time, erlang:now/0 also produces unique and strictly monotonically increasing values. To detach this functionality from time measurements, we have introduced erlang:unique_integer().

How to Work with the New API

Previously erlang:now/0 was the only option for doing many things. This section deals with some things that erlang:now/0 can be used for, and how you use the new API.

Retrieve Erlang System Time

Use erlang:now/0 to retrieve the current Erlang system time.

Use erlang:system_time/1 to retrieve the current Erlang system time on the time unit of your choice.

If you want the same format as returned by erlang:now/0, use erlang:timestamp/0.

Measure Elapsed Time

Take time stamps with erlang:now/0 and calculate the difference in time with timer:now_diff/2.

Take time stamps with erlang:monotonic_time/0 and calculate the time difference using ordinary subtraction. The result is in native time unit. If you want to convert the result to another time unit, you can use erlang:convert_time_unit/3.

An easier way to do this is to use erlang:monotonic_time/1 with the desired time unit. However, you can then lose accuracy and precision.

Determine Order of Events

Determine the order of events by saving a time stamp with erlang:now/0 when the event occurs.

Determine the order of events by saving the integer returned by erlang:unique_integer([monotonic]) when the event occurs. These integers are strictly monotonically ordered on current runtime system instance corresponding to creation time.

Determine Order of Events with Time of the Event

Determine the order of events by saving a time stamp with erlang:now/0 when the event occurs.

Determine the order of events by saving a tuple containing monotonic time and a strictly monotonically increasing integer as follows:

Time = erlang:monotonic_time(), UMI = erlang:unique_integer([monotonic]), EventTag = {Time, UMI}

These tuples are strictly monotonically ordered on the current runtime system instance according to creation time. It is important that the monotonic time is in the first element (the most significant element when comparing two-tuples). Using the monotonic time in the tuples, you can calculate time between events.

If you are interested in Erlang system time at the time when the event occurred, you can also save the time offset before or after saving the events using erlang:time_offset/0. Erlang monotonic time added with the time offset corresponds to Erlang system time.

If you are executing in a mode where time offset can change, and you want to get the actual Erlang system time when the event occurred, you can save the time offset as a third element in the tuple (the least significant element when comparing three-tuples).

Create a Unique Name

Use the values returned from erlang:now/0 to create a name unique on the current runtime system instance.

Use the value returned from erlang:unique_integer/0 to create a name unique on the current runtime system instance. If you only want positive integers, you can use erlang:unique_integer([positive]).

Seed Random Number Generation with a Unique Value

Seed random number generation using erlang:now().

Seed random number generation using a combination of erlang:monotonic_time(), erlang:time_offset(), erlang:unique_integer(), and other functionality.

To sum up this section: Do not use erlang:now/0.

Support of Both New and Old OTP Releases

It can be required that your code must run on a variety of OTP installations of different OTP releases. If so, you cannot use the new API out of the box, as it will not be available on releases before OTP 18. The solution is not to avoid using the new API, as your code would then not benefit from the scalability and accuracy improvements made. Instead, use the new API when available, and fall back on erlang:now/0 when the new API is unavailable.

Fortunately most of the new API can easily be implemented using existing primitives, except for:

erlang:system_info(start_time) erlang:system_info(end_time) erlang:system_info(os_monotonic_time_source) erlang:system_info(os_system_time_source))

By wrapping the API with functions that fall back on erlang:now/0 when the new API is unavailable, and using these wrappers instead of using the API directly, the problem is solved. These wrappers can, for example, be implemented as in $ERL_TOP/erts/example/time_compat.erl.