# How Is Server-Timing used on the web?

**URL:** <https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034>\
**Category:** Analysis\
**Created:** [July 27, 2017, 9:26pm UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034 "2017-07-27T21:26:33Z")\
**Posts on this page:** 18\
**Page:** 1

<div class="post-metadata">

**Author:** ![paulcalvano](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/paulcalvano/32/1601_2.png) [@paulcalvano](https://discuss.httparchive.org/u/paulcalvano)\
**Post date:** [July 27, 2017, 9:26pm UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/1 "2017-07-27T21:26:33Z")

</div>

I was curious to see where[Server-Timing](https://w3c.github.io/server-timing/) was implemented on the web, so I started searching the HTTP Archive for sites using it. Interestingly enough, there were no sites in the HTTP Archive that had Server-Timing response headers before 3/1/2017. Since then it’s usage has been gradually increasing each month. As of July 2017, there are 72 sites and 352 HTTP responses containing Server-Timing headers.

Here’s a query that I wrote to find a list of sites with Server-Timing headers. [You can run it on BigQuery here](https://bigquery.cloud.google.com/savedquery/264421607889:45510569d166427594842ad55582f2f2) :

```sql
    SELECT p.url, p.rank, count(*) as freq
    FROM 
      ( SELECT pageid, url, rank 
        FROM [httparchive:runs.latest_pages] 
      ) p
    JOIN  
      ( SELECT pageid,url,respOtherHeaders 
        FROM [httparchive:runs.latest_requests] 
        WHERE LOWER(respOtherHeaders) CONTAINS "server-timing"
      ) r
    ON p.pageid = r.pageid
    GROUP BY p.url, p.rank
    ORDER BY freq DESC

```

![image](https://canada1.discourse-cdn.com/flex035/uploads/httparchive/original/1X/7b93438aab12108a1d255402407ef9b977abc7cd.png)

Next I was curious to see the actual Server-Timing headers to understand how they are being used. I wrote this query to strip out the server-timing headers from the respOtherHeaders column. The query has a lot of string manipulations to extract the text. I’ve added some comments to the query to help make it easier to follow -

```sql
    SELECT HOST(url), status, mimeType,
    IF (
        -- In respOtherHeaders, remove everything to the left of the server-timing header. Is there a ; in the remaining string?
        RIGHT(respOtherHeaders,LENGTH(respOtherHeaders)-INSTR(LOWER(respOtherHeaders),"server-timing")+1) CONTAINS ";",
        -- Then strip out everything to the left of the server-timing header and everything to the right of the ";" character.
        LEFT(
            RIGHT(respOtherHeaders,
                LENGTH(respOtherHeaders)-INSTR(LOWER(respOtherHeaders),"server-timing")+1),
            INSTR(
                RIGHT(respOtherHeaders,
                    LENGTH(respOtherHeaders)-INSTR(LOWER(respOtherHeaders),
                    "server-timing")+1),
                ";")
            ), 
        -- If there are no ; chars, then the respOtherHeaders column ends with server-timing. Remove everything to the left of server-timing
        RIGHT(respOtherHeaders,
            LENGTH(respOtherHeaders)-INSTR(LOWER(respOtherHeaders),
            "server-timing")+1)
    ) as ServerTimingHeader
    FROM [httparchive:runs.latest_requests] 
    WHERE LOWER(respOtherHeaders) CONTAINS "server-timing"

```

[You can see it on BigQuery here](https://bigquery.cloud.google.com/savedquery/264421607889:a85d7b23c59849e0bd2827c89a9c0c64)

 ![image](https://canada1.discourse-cdn.com/flex035/uploads/httparchive/original/1X/1ed880792d062a7f371ba7e43ea722c5f93ca378.png)

Looking at the Server-Timing headers, I can see 3 different types of data being sent in these headers:

**Timing Metrics** - such as application time, gateway time, image resize time, etc  
Server-Timing = app=133.99219512939;  
Server-Timing = gateway-time=25;  
Server-Timing = resizer=0.015;  
server-timing = view=2.188, db=0.000, total=61.743, x-runtime = 0.069661, x-ua-compatible = ie=edge,chrome=1, x-frame-options = SAMEORIGIN  
Server-Timing = app=53.393840789795;

**Caching Information**  
server-timing = hit, cf-cache-status = HIT  
server-timing = cold, cf-cache-status = MISS  
server-timing = hit, cf-cache-status = HIT  
server-timing = cold, cf-cache-status = MISS

**Informational**  
server-timing = ibs\_5318e9209631=1  
server-timing = ibs\_1964fd84c9cea3=1  
server-timing = ibs\_1880bf4b0f6311=1

Looking at all of the data in the 7/1/2017 tables, I can see that the current examples in the wild are split between these 3 types of data, although the caching ones are mostly on images, the informational ones are set on JS and responses w/ no content (ie, 204s -likely beacon responses), and the actual timing metrics are spread across a wide range of content.

 ![image](https://canada1.discourse-cdn.com/flex035/uploads/httparchive/original/1X/fc32107cf45823ee8c50a3652a0ebb826fc008ac.png)

It’ll be interesting to see how this evolves as Server-Timing adoption increases over time.

---

<div class="post-metadata">

**Author:** ![rviscomi](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/rviscomi/32/835_2.png) [@rviscomi](https://discuss.httparchive.org/u/rviscomi)\
**Post date:** [July 27, 2017, 9:49pm UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/2 "2017-07-27T21:49:07Z")

</div>

Great stuff, thanks for sharing!

---

<div class="post-metadata">

**Author:** ![cvazac](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/cvazac/32/796_2.png) [@cvazac](https://discuss.httparchive.org/u/cvazac)\
**Post date:** [July 28, 2017, 1:24am UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/3 "2017-07-28T01:24:16Z")

</div>

I think `cf-cache-status` is a different header.

The numbers show that by an overwhelming margin it’s being used more to share meta data than timing. I’m surprised that it’s not closer.

Thanks for doing the legwork!

---

<div class="post-metadata">

**Author:** ![paulcalvano](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/paulcalvano/32/1601_2.png) [@paulcalvano](https://discuss.httparchive.org/u/paulcalvano)\
**Post date:** [July 28, 2017, 1:55am UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/4 "2017-07-28T01:55:49Z")

</div>

It looks like some of the headers in the respOtherHeaders field are `;` delimited while others are `,` delimited. That may explain how `cf-cache-status` snuck in there.

Also @cvazac asked me offline what the trend line looked like for sites using Server-Timing.

 ![image](https://canada1.discourse-cdn.com/flex035/uploads/httparchive/original/1X/487f11112b24d6f19734e21521e70ac4fadfcf4e.png)

The query for that was:

```sql
    SELECT label, count(distinct p.url)
    FROM 
      ( SELECT label, pageid, url, rank 
        FROM [httparchive:runs.2017_07_01_pages],
             [httparchive:runs.2017_06_15_pages],
             [httparchive:runs.2017_06_01_pages],
             [httparchive:runs.2017_05_15_pages],
             [httparchive:runs.2017_05_01_pages],
             [httparchive:runs.2017_04_15_pages],
             [httparchive:runs.2017_04_01_pages],
             [httparchive:runs.2017_03_15_pages],
             [httparchive:runs.2017_03_01_pages],
             [httparchive:runs.2017_02_15_pages],
             [httparchive:runs.2017_02_01_pages] 
      ) p
    JOIN  
      ( SELECT pageid,url,respOtherHeaders 
        FROM [httparchive:runs.2017_07_01_requests],
             [httparchive:runs.2017_06_15_requests],
             [httparchive:runs.2017_06_01_requests],
             [httparchive:runs.2017_05_15_requests],
             [httparchive:runs.2017_05_01_requests],      
             [httparchive:runs.2017_04_15_requests],
             [httparchive:runs.2017_04_01_requests],
             [httparchive:runs.2017_03_15_requests],
             [httparchive:runs.2017_03_01_requests],
             [httparchive:runs.2017_02_15_requests],
             [httparchive:runs.2017_02_01_requests] 
        WHERE LOWER(respOtherHeaders) CONTAINS "server-timing"
      ) r
    ON p.pageid = r.pageid
    GROUP BY label

```

---

<div class="post-metadata">

**Author:** ![simevidas](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/simevidas/32/661_2.png) [@simevidas](https://discuss.httparchive.org/u/simevidas)\
**Post date:** [July 28, 2017, 9:23pm UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/5 "2017-07-28T21:23:21Z")

</div>

@cvazac @paulcalvano `cf-cache-status` is a header that Cloudflare includes in HTTP responses when proxy CDN mode is on. However, notice that in the data above, it is always preceded by a `hit` or `cold` value. This cannot be a coincidence. It seems that the websites are using `server-timing` to make the value of `cf-cache-status` (HIT or MISS) available to JavaScript in their client-side web app.

Although I don’t know how they’re doing it. Afaik, if Cloudflare has a HIT, the request doesn’t even reach the website’s (origin) server, so how can the website’s server set the `server-timing` header in that case?

---

<div class="post-metadata">

**Author:** ![paulcalvano](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/paulcalvano/32/1601_2.png) [@paulcalvano](https://discuss.httparchive.org/u/paulcalvano)\
**Post date:** [July 28, 2017, 10:28pm UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/6 "2017-07-28T22:28:20Z")

</div>

There were a few examples where the `cf-cache-status` header did not match up to the server timing `hit` or `cold` value. For example

```
server-timing = miss, cf-cache-status = HIT

```

I suspect the site that implemented this was tracking their origin cache’s status in a server-timing header. If that is the case, and multiple cache tiers are utilized (origin + CDN) then server-timing header set by the origin server would likely be swapped into the CDN cache and become more of an informational metric (perhaps useful for debugging) rather than a real time cache status.

---

<div class="post-metadata">

**Author:** ![igrigorik](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/igrigorik/32/925_2.png) [@igrigorik](https://discuss.httparchive.org/u/igrigorik)\
**Post date:** [July 30, 2017, 2:59am UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/7 "2017-07-30T02:59:02Z")

</div>

@paulcalvano I think we’re hitting some false positives in above query: cf-cache-status, timing-allow-origin, acao - these are all standalone header fields. Not sure what exactly is going wrong in above query, but we can probably avoid a bunch of edge cases here by looking at the HAR data, which contains the JSON-encoded response headers.

```sql
CREATE TEMPORARY FUNCTION getHeaders(arr STRING)
RETURNS STRING
LANGUAGE js AS """
  try {
    var headers = JSON.parse(arr);
    var st = headers.find(function(e) { 
      return e['name'].toLowerCase() === 'server-timing'
    });
    return st ? st['value'] : null;
  } catch (e) {
    return [];
  }
""";

SELECT * FROM (
  SELECT 
    url, getHeaders(JSON_EXTRACT(payload, '$.response.headers')) as st
  FROM httparchive.har.2017_07_01_chrome_requests 
) WHERE st != ""

```

Some examples Server-Timing values, based on above:

```sql
ibs_d60693ef000eb=1
ibs_33968ba999d7f=1
"ssr=0.161,fetch_data=0.098"
ibs_e01814003491b=1
ibs_14b97674ae49a5=1
_v1_nearest_region=0.043
"view=0.000, db=2.848, total=32.724"
"view=0.000, db=2.140, total=24.670"
ibs_10fcc10c812fdb=1
ibs_10111b2d59e017=1
ibs_5c6f3fa2213b1=1
"view=129.949, db=183.955, total=361.542"
ibs_394d30379c8bd=1
"view=0.190, db=2.409, total=20.650"
"view=0.243, db=1.935, total=206.689"
"Bootstrap=27, SQL=22, Process=150, Total=177"

```

---

<div class="post-metadata">

**Author:** ![paulcalvano](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/paulcalvano/32/1601_2.png) [@paulcalvano](https://discuss.httparchive.org/u/paulcalvano)\
**Post date:** [July 30, 2017, 6:19pm UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/8 "2017-07-30T18:19:36Z")

</div>

Thanks @igrigorik. I hadn’t tried querying the HAR files for this type of information before. Very cool!

It does look like there are still some discrepencies between the examples that both queries found.

The additional headers were added to my output because of how the respOtherHeaders column is formatted with comma delimiters. If I would have attempted to extract a server-timing header by a comma, I’d miss some server-timing data. When I attempted to extract based on a semicolon I had extra data.

If we take a look at `pageid=78751326 & requestid=532345152` for example, I can see that respOtherHeaders contains the following:

`status = 200, x-content-type-options = nosniff, content-security-policy = upgrade-insecure-requests, server-timing = cold, cf-cache-status = MISS, strict-transport-security = max-age=5184000; includeSubDomains, access-control-allow-origin = *, link = <http://www.accommodationforstudents.com//images/properties/150118_Main.jpg>; rel="canonical", cf-ray = 379c546c3e641e8f-SJC, x-frame-options = SAMEORIGIN`

I was able to run your query, but do not see this example and others like it in the output. I do see it in the json payload for `page="http://www.accommodationforstudents.com/" and url = "https://img.gs/vvjnmlqfld/200/http://www.accommodationforstudents.com//images/properties/150118_Main.jpg"` , and since name=‘server-timing’ and the value is not “”, I’m not sure why it wasn’t caught…

![image](https://canada1.discourse-cdn.com/flex035/uploads/httparchive/original/1X/da9bf60e3ff5c7cac87ca93f42ead0e6afca4a8a.png)

---

<div class="post-metadata">

**Author:** ![igrigorik](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/igrigorik/32/925_2.png) [@igrigorik](https://discuss.httparchive.org/u/igrigorik)\
**Post date:** [July 30, 2017, 7:32pm UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/9 "2017-07-30T19:32:29Z")

</div>

Interesting. Zeroing in on that record in particular…

```sql
CREATE TEMPORARY FUNCTION getHeaders(arr STRING)
RETURNS STRING
LANGUAGE js AS """
  try {
    var headers = JSON.parse(arr);
    var st = headers.find(function(e) { 
      return e['name'].toLowerCase() === 'server-timing'
    });
    return st ? st['value'] : null;
  } catch (e) {
    return [e];
  }
""";

  SELECT 
    url, 
    JSON_EXTRACT(payload, '$.response.headers') as st,
    getHeaders(JSON_EXTRACT(payload, '$.response.headers'))
  FROM httparchive.har.2017_07_01_chrome_requests 
  where page = "http://www.accommodationforstudents.com/" 
  AND url = "https://img.gs/vvjnmlqfld/200/http://www.accommodationforstudents.com//images/properties/150118_Main.jpg"

```

If you run that, we get a JSON exception:

> SyntaxError: Unexpected token in JSON at position 688

The culprit is:

> {“name”:“etag”,“value”:““im-vGx8npx7m3l64n42Wy””}

Chrome’s export as HAR for that record returns correctly encoded output:

> {  
> “name”: “etag”,  
> “value”: “"im-vGx8npx7m3l64n42Wy"”  
> },

It appears that we have an escaping issue somewhere in our pipeline; the plot thickens. 😉

@rviscomi @patmeenan I’m trying to hunt down the source HAR, but not having a lot of luck. Any ideas what might be going on here? I suspect the HAR file is correct, but perhaps the BQ encode logic is not applying correct escaping?

---

<div class="post-metadata">

**Author:** ![paulcalvano](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/paulcalvano/32/1601_2.png) [@paulcalvano](https://discuss.httparchive.org/u/paulcalvano)\
**Post date:** [July 30, 2017, 11:38pm UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/10 "2017-07-30T23:38:14Z")

</div>

Ooh - a mystery :). The requestid for this example, matches up with the page id `78751326`. So a search for

```sql
SELECT pageid, wptid, wptrun, url
FROM [httparchive:runs.2017_07_01_pages] 
WHERE pageid = 78751326

```

tells me that we’re looking at run #3 from test ID `170701_15_5RTJ`. So this should be where you can download the HAR file from - [http://httparchive.webpagetest.org/result/170701\_15\_5RTJ/3/details/](http://httparchive.webpagetest.org/result/170701_15_5RTJ/3/details/).

A quick look at the HAR file shows that the value in the HAR is escaped (pasted below w/ spaces removed for readability)

```sql
 {
   "name": "etag",
   "value": "\"im-vGx8npx7m3l64n42Wy\""
},

```

The quotes are 0x22 chars and appear to be escaped with a backslash. So it looks like this is not an issue with the HAR file:

![image](https://canada1.discourse-cdn.com/flex035/uploads/httparchive/original/1X/8747087fbf2e64b1961fc98deed69c110f31ad5f.png)

---

<div class="post-metadata">

**Author:** ![patmeenan](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/patmeenan/32/605_2.png) [@patmeenan](https://discuss.httparchive.org/u/patmeenan)\
**Post date:** [July 31, 2017, 1:27pm UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/11 "2017-07-31T13:27:43Z")

</div>

Thanks for digging up the test ID. The HAR file that went into the BQ pipeline is [here (chrome-Jul\_1\_2017/170701\_15\_5RTJ.har.gz)](https://storage.googleapis.com/httparchive/chrome-Jul_1_2017/170701_15_5RTJ.har.gz). It is basically the same, just not pretty-printed in the raw form:

```sql
{"name":"etag","value":"\"im-vGx8npx7m3l64n42Wy\""}

```

Definitely looks like it is downstream in BQ somewhere.

---

<div class="post-metadata">

**Author:** ![rviscomi](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/rviscomi/32/835_2.png) [@rviscomi](https://discuss.httparchive.org/u/rviscomi)\
**Post date:** [August 1, 2017, 2:07am UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/12 "2017-08-01T02:07:13Z")

</div>

> [@igrigorik](#):
>
> If you run that, we get a JSON exception

I ran it but not seeing a JSON exception. Am I missing something?

 ![image](https://canada1.discourse-cdn.com/flex035/uploads/httparchive/original/1X/4ca1595054e835750e6be1899e1c0f7cac14b96e.png)

---

<div class="post-metadata">

**Author:** ![paulcalvano](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/paulcalvano/32/1601_2.png) [@paulcalvano](https://discuss.httparchive.org/u/paulcalvano)\
**Post date:** [August 1, 2017, 2:20am UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/13 "2017-08-01T02:20:25Z")

</div>

Scroll all the way to the right. The `getHeaders(JSON_EXTRACT(payload, '$.response.headers'))` function call is what is returning the error -

 ![image](https://canada1.discourse-cdn.com/flex035/uploads/httparchive/original/1X/eef7928094754b91b38280b99d1764a675512010.png)

---

<div class="post-metadata">

**Author:** ![paulcalvano](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/paulcalvano/32/1601_2.png) [@paulcalvano](https://discuss.httparchive.org/u/paulcalvano)\
**Post date:** [August 1, 2017, 2:35am UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/14 "2017-08-01T02:35:39Z")

</div>

Also, running the following query to understand the scope of the JSON parsing error (outside of server-timing data) -

```sql
CREATE TEMPORARY FUNCTION getHeaders(arr STRING)
RETURNS STRING
LANGUAGE js AS """
  try {
    var headers = JSON.parse(arr);
    var st = headers.find(function(e) { 
      return e['name'].toLowerCase() === 'date'
    });
    return "No Error";
  } catch (e) {
    return "Error";
  }
""";

  SELECT getHeaders(JSON_EXTRACT(payload, '$.response.headers')) as header, count(*)
  FROM httparchive.har.2017_07_01_chrome_requests   
  GROUP BY header

```

Shows that there are a lot of errors parsing something as simple as a date out of the imported HAR files. Since it’s fairly common for etags to be quoted, I imagine that this bug would affect other queries that are run against the json in the HAR.

![image](https://canada1.discourse-cdn.com/flex035/uploads/httparchive/original/1X/51719b161cb4c3a013892b9eddee55a6049aa20c.png)

---

<div class="post-metadata">

**Author:** ![igrigorik](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/igrigorik/32/925_2.png) [@igrigorik](https://discuss.httparchive.org/u/igrigorik)\
**Post date:** [August 1, 2017, 2:52am UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/15 "2017-08-01T02:52:36Z")

</div>

@paulcalvano that’s a lot of errors 😱 … on the bright side, there’s probably a common fix for all of them.

---

<div class="post-metadata">

**Author:** ![rviscomi](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/rviscomi/32/835_2.png) [@rviscomi](https://discuss.httparchive.org/u/rviscomi)\
**Post date:** [August 3, 2017, 6:29pm UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/16 "2017-08-03T18:29:42Z")

</div>

The issue is that we’re double JSON parsing the payload in the query. The etag value started correctly encoded as `\"foo\"` but after the first parse it became `"foo"` and failed on the second parse.

Rather than using JSON\_EXTRACT to narrow down the payload to an array and a UDF to parse the array, we should just pass the entire payload to the UDF and do all of the parsing there.

 ![image](https://canada1.discourse-cdn.com/flex035/uploads/httparchive/original/1X/dae0db5ef3f781bafcacdda84a37d86c9eb60f54.png)

---

<div class="post-metadata">

**Author:** ![rviscomi](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/rviscomi/32/835_2.png) [@rviscomi](https://discuss.httparchive.org/u/rviscomi)\
**Post date:** [August 3, 2017, 7:18pm UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/17 "2017-08-03T19:18:53Z")

</div>

Modified @igrigorik’s query from earlier:

```sql
CREATE TEMPORARY FUNCTION getHeaders(payload STRING)
RETURNS STRING
LANGUAGE js AS """
  try {
    var $ = JSON.parse(payload);
    var headers = $.response.headers;
    var st = headers.find(function(e) { 
      return e['name'].toLowerCase() === 'server-timing'
    });
    return st['value'];
  } catch (e) {
    return '';
  }
""";

SELECT
  *
FROM (
  SELECT
    url,
    getHeaders(payload) AS st
  FROM
    httparchive.har.2017_07_01_chrome_requests )
WHERE
  st != ""

```

[BigQuery](https://bigquery.cloud.google.com/savedquery/226352634162:d7f220b11a164c8cb0be94c3f667317e)

There are 357 requests with Server-Timing headers. I also modified that a bit to count the number of distinct pages and got 72, which is also what @paulcalvano got in the timeseries chart.

```sql
SELECT
  page
FROM (
  SELECT
    page,
    url,
    getHeaders(payload) AS st
  FROM
    httparchive.har.2017_07_01_chrome_requests )
WHERE
  st != ""
GROUP BY
  page

```

---

<div class="post-metadata">

**Author:** ![igrigorik](https://yyz1.discourse-cdn.com/flex035/user_avatar/discuss.httparchive.org/igrigorik/32/925_2.png) [@igrigorik](https://discuss.httparchive.org/u/igrigorik)\
**Post date:** [August 3, 2017, 7:31pm UTC](https://discuss.httparchive.org/t/how-is-server-timing-used-on-the-web/1034/18 "2017-08-03T19:31:28Z")

</div>

Ahh, doh! Kudos for tracking this down Rick.