A Simple Performance Comparison of HTTPS, SPDY and HTTP/2

calendarJanuary 16, 2015 in Firefox , HTTP/2 , HTTPS , HttpWatch , SPDY , SSL

Firefox 35 was released this week and became the first browser to enable support for the HTTP/2 protocol by default.

The HTTP/2 specification has not been finalised so Firefox actually enabled the Draft 14 version of HTTP/2 but little is expected to change in the final draft. Google is now supporting HTTP/2 draft 14 on its web servers alongside the SPDY protocol giving us a chance to compare the performance of raw HTTPS, SPDY and HTTP/2 on the same web page.

We also updated HttpWatch this week so that it supports HTTP/2 within Firefox. It has new columns to display information about the protocols being used by each request:

New HTTP/2 Columns

The Performance Comparison

The performance test used HttpWatch with Firefox to run a series of simple page load tests against the Google UK home page using the three protocols:

  • Raw HTTPS
  • SPDY/3.1
  • HTTP/2

We switched between the protocols by enabling and disabling the following entries in Firefox’s about:config page:

Controlling FIrefox Protocol Support

Each test was performed in a fresh instance of Firefox with an empty browser cache. Although this testing was simplistic and only used a simple page it does highlight some important differences between the protocols.

Test #1 – Size of Request and Response Headers

Most sites already use compression when downloading textual content as it provides a significant performance benefit. Unfortunately, HTTP/1.1 doesn’t support the compression of HTTP headers that are attached to every HTTP request and response. SPDY and subsequently HTTP/2 were designed to address this shortcoming using different types of compression.

SPDY uses the general purpose DEFLATE algorithm whereas HTTP/2 uses HPACK that was specifically designed to compress headers. It uses predefined tokens, dynamic tables and Huffman compression.

It’s possible to see the difference in the generated header sizes by looking at a request that has no content. On the Google UK home page there is beacon request that returns no content (a 204 response). These screen shots from HttpWatch show the size of the request headers in the ‘Sent’ column and the size of the response headers in the ‘Received’ column:

HTTPS header sizesSPDY header sizes

HTTP/2 header sizes

WINNER: HTTP/2

HTTP/2 has significantly smaller header sizes due to its use of the HPACK algorithm.

Test #2: Size of Response Message

The response message from the server is made up of the response headers and the encoded response content. Given the fact that HTTP/2 creates the smallest headers shouldn’t it always have the smallest response message?

In HttpWatch this seems to be the case for image resources:

HTTPS Image Response Size SPDY Image Response SizeHTTP/2 Image Response Size

However, for textual resources SPDY consistently has smaller response messages even though its headers would be larger than HTTP/2:

HTTPS Text Response Size SPDY Text Response Size HTTP/2 Text Response Size

The reason for this is the optional padding bytes that can be added to the HTTP/2 DATA frame. HttpWatch doesn’t currently show the padding but in our debug logs we can see that the Google servers add padding to the data frames of textual resources. The reason given in the HTTP/2 spec for using padding is:

Padding can be used to obscure the exact size of frame content, and is provided to mitigate specific attacks within HTTP. For example, attacks where compressed content includes both attacker-controlled plaintext and secret data (see for example, [BREACH]).

Padding isn’t used for image resources because they already have a compressed format that will not contain attacker controlled plain text.

WINNER: SPDY

The larger response bodies seen on the Google servers are due to the use of padding in data frames. Although, HTTP/2 produces larger responses than SPDY its encrypted connections could potentially be more secure. This may be an area where tuning can be performed in a trade off between security and performance.

Test #3 : Number of TCP Connects and SSL Handshakes Required During Page Load

Browsers achieved a performance boost in HTTP/1.1 by increasing the maximum number of connections per host name from two to six or more. This allowed greater concurrency during the download of a page at the cost of extra TCP connections and SSL handshakes . Increasing concurrency allows the bandwidth of the network to be used more effectively because it reduces the blocking of requests.

SPDY and HTTP/2 support concurrency on a single TCP and SSL connection by using multiplexing to allow more than one request at a time to send and receive data on a single connection.

By adding the ‘Connect’ and ‘SSL Handshake’ timings as a columns in HttpWatch you can see that SPDY:

SPDY Connections

And HTTP/2:

HTTP/2 Connections

Only create connections for different host names. Whereas, raw HTTPS may create more than one connection per host name to improve concurrency:

HTTPS Connections

EQUAL WINNERS: SPDY & HTTP/2

The multiplexing support added in SPDY and HTTP/2 reduces the number of connections that have to be setup to download a page. As a side benefit web servers will not have to maintain as many active TCP connections when HTTP/2 usage becomes more widespread.

Test #4: Page Load Time

The ‘Page Load’ event in HttpWatch shows when the page was fully downloaded and available for use. In most cases this a reasonable measure of the speed of the page as seen by a visitor to a web site.

The screeen shots below show the Page Load time for each of the three protocols:HTTPS Page Load

SPDY Page Load HTTP/2 Page Load

WINNER HTTP/2

The Page Load timing was worse for raw HTTPS probably due to the lack of header compression and the additional TCP connections and SSL Handshakes required. For more complex pages the speed advantage of SPDY and HTTP/2 should be even greater.

We also found that HTTP/2 was consistently faster than SPDY even though its response messages were often larger. The advantage was probably due to the smaller GET request messages produced by HPACK compression. Our network connection to the internet, like many others, is asymmetric in nature – the network upload speed is less than the download speed. This means that any saving in uploaded data has much more impact than the equivalent saving in downloaded data.

Conclusion

HTTP/2 is likely to provide significant performance advantages compared to raw HTTPS and even SPDY. However the use of padding in response messages is an area of potential concern where there could be a trade-off between performance and security.

You can check for HTTP/2 using our new SSL test tool SSLRobot . It will also look for potential issues with the certificates and TLS/SSL configuration used by your site. Try it now for free!

 

HttpWatch 9.2: SSL handshake and Protocol Information in Firefox

calendarFebruary 14, 2014 in Firefox , HTTPS , HttpWatch , SSL

HttpWatch 9.2 is now available for download and brings the level of SSL reporting in Firefox up to the same level as the plugin for IE and the iOS app.

SSL handshake timings are now displayed in Firefox:

SSL Handshake Timing

and in-depth information about the SSL protocol used by each connection:

SSL Information

We’ve also made some other SSL related improvements that are available in the Firefox/IE plugins and the HttpWatch Studio log file viewer. The first is that SSL information can now be added as columns in the main request grid:

SSL Columns

v92_ssl_columns_grid

There’s also a new warning that can be used to highlight HTTPS connections that have potential vulnerabilities:

SSL Warning

You can check SSL/TLS configuration our new SSL test tool SSLRobot . It will also look for potential issues with the certificates, ciphers and protocols used by your site. Try it now for free!

 

Top 7 Myths about HTTPS

calendarJanuary 28, 2011 in Firefox , HTTPS , HttpWatch

Myth #7 – HTTPS Never Caches

People often claim that HTTPS content is never cached by the browser; perhaps because that seems like a sensible idea in terms of security. In reality, HTTPS caching is controllable with response headers just like HTTP.

Eric Lawrence explains this succinctly in his IEInternals blog:

It comes as a surprise to many that by-default, all versions of Internet Explorer will cache HTTPS content so long as the caching headers allow it. If a resource is sent with a Cache-Control: max-age=600 directive, for instance, IE will cache the resource for ten minutes. The use of HTTPS alone has no impact on whether or not IE decides to cache a resource. (Non-IE browsers may have different default behavior for caching of HTTPS content, depending on which version you’re using, so I won’t be talking about them.)

The slight caveat is that Firefox will only cache HTTPS resources in memory by default. If you want persistent caching to disk you’ll need to add the Cache-Control: Public response header.

This screenshot shows the contents of the Firefox disk cache and the Cache-Control: Public response header in HttpWatch:

Myth #6 – SSL Certificates are Expensive

If you shop around you can find SSL certificates for about $ 10 a year or roughly the same cost as the registration of a .com domain for a year.

(UPDATE: you can get domain validated SSL certificates for free. See comment #1)

The cheapest certificates don’t have the level of company verification provided by the more expensive alternatives but they do work with nearly all mainstream browsers.

Myth #5 – Each HTTPS Site Needs its Own Public IP Address

With the pool of IPv4 addresses running low this is a valid concern and it’s true that only one SSL certificate can be installed on single IP address. However, if you have a wildcard SSL certificate (from about $ 125 yr) you can have as many sub-domains as you like on a single IP address. For example, we run https://www.httpwatch.com, http://www.httpwatch.com and https://store.httpwatch.com on the same public IP address:

On IIS 7 there is a trick though to making this work. After adding a certificate you need to find it and rename it in the certificate manager so that the name starts with a *. If you don’t do this you cannot edit the hostname field for an HTTPS binding:

UPDATE: UCC (Unified Communications Certificate) supports multiple domains in a single SSL certificate and can be used where you need to secure several sites that are not all sub-domains.

UPDATE #2: SNI (Server Name Indication) allows multiple certificates for different domains to be hosted on the same IP address. On the server side it’s supported by Apache and Nginx, but not IIS. On the client it’s supported by IE 7+, Firefox 2.0+, Chrome 6+, Safari 2.1+ and Opera 8.0+.  See comment #4 and comment #5.

Myth #4 – New SSL Certificates Have to be Purchased When Moving Servers or Running Multiple Servers

Buying an SSL certificate involves:

  1. Creating a CSR (SSL Certificate Signing Request) on your web server
  2. Purchasing the SSL certificate using the CSR
  3. Installing the SSL certificate by completing the CSR process

These steps are designed to ensure that the certificate is safely transferred to the web server and prevents anyone from using the certificate if they intercept any emails or downloads containing the certificate in step 2).

The result is that you cannot just use the files from step 2) on another web server. If you want to do that you’ll need to export the certificate in other format.

In IIS you can create a transferrable .pfx file that is protected by a password:

This file can be imported onto other web servers by supplying the password again.

Myth #3 – HTTPS is Too Slow

Using HTTPS isn’t going to make your site faster (actually it can – see below) but the overhead is mostly avoidable by following the tips in our HTTPS Performance Tuning blog post.

The amount of CPU resource required to encrypt the data can be reduced by compressing textual content and is usually not a significant on servers with modern CPUs.

Extra TCP level round-trips are required to setup HTTPS connections and some additional bytes have to be sent and received. However, you can see in HttpWatch that this overhead is small once the HTTPS connection has been made:

The initial visit to an HTTPS site is somewhat slower than HTTP due to the longer connection times required to setup SSL. Here’s a time chart of the page load for an HTTP site recorded in HttpWatch:

And here’s the same site accessed over HTTPS:

The longer connection times caused the initial page load to be about 10% slower. However, once the browser has active keep-alive HTTPS connections a subsequent refresh of the page shows very little difference between HTTP and HTTPS.

First, the page refresh with HTTP:

and then with HTTPS:

It’s possible that some users may even find that the HTTPS version of a web site is faster than HTTP. This can happen if they sit behind a coporate HTTP proxy that normal intercepts, examines and records web traffic. An HTTPS connection will often just be forwarded as a simple TCP connection through the proxy because HTTPS traffic cannot be intercepted. It’s this bypassing that can lead to improved performance.

UPDATE: A blog post by F5 challenges the claim the CPU overhead of SSL is no longer significant, but most of their arguments are refuted in this follow up.

Myth #2 – Anything can go in Cookies and Query Strings with HTTPS

Although, a hacker cannot intercept a user’s HTTPS traffic on the network and read their cookie or query string values directly, you still need to ensure that their values can’t be easily predicted.

For example, one of the early UK banking sites used simple counter based numeric values for the session id:

A hacker could use a dummy account to see how this cookie worked and find a recent value. They could then try manipulating the cookie value in their own browser to hi-jack other sessions with nearby session id values.

Query string values are also protected on the network by HTTPS but they can still leak their values in other ways. For more details see How Secure Are Query Strings Over HTTPS .

Myth #1 – My Site Only Needs HTTPS for the Login Page

This is a commonly held view. The theory being that HTTPS will protect the user’s password during login but HTTPS is not needed after that.

The recently released Firesheep add-on for Firefox demonstrated the fallacy of this approach and how easy it is to hi-jack someone’s else session on sites like Twitter and Facebook.

The free public WiFi in a coffee shop is an ideal environment for session hi-jacking because:

  • The WiFi network doesn’t normally use encryption so it’s very easy to monitor all traffic
  • The WiFi network probably uses NAT through a single IP address to access the internet. This means that a highjacked session appears to come from the same network address as the original login

There are lots of examples of this approach to security. For example, by default the Twitter signin page uses HTTPS but it then switches to HTTP after setting up the session level cookies:

HttpWatch warns that these cookies were setup on HTTPS but the Secure flag wasn’t used to prevent them being used with HTTP:

Potentially someone in a coffee shop with Firesheep could intercept your twitter session cookies and then hi-jack your session to start tweeting on your behalf.

You can check SSL/TLS configuration our new SSL test tool SSLRobot . It will also look for potential issues with the certificates, ciphers and protocols used by your site. Try it now for free!

Ready to get started? TRY FOR FREE Buy Now