• Performance Analytics – See inside your live app
• Real User Monitoring – Optimize for humans, not bots
• Server Monitoring – Discover how server affect apps
• SQL/NoSQL Performance – Map database and app performance
• Web Transaction Tracing – Zero in on problems fast
• Proactive Notifications – Be the first to know
• Deployment History – What changed and when?
• Availability Monitoring – Uptime alerting and reporting
• Map – Quickly map architecture
• Scalability Analysis – But does it scale?
• Java Profiling – Timing issues, locks and more

Unlike our existing monitoring systems that monitor at the service level, New Relic allows you to monitor details within your application stack.

Try out New Relic free for 14 days.

If you are looking to test out New Relic sign up with them first. Then just supply us with your your license key, and we’ll do the rest.

The problem is management of the kernel. An outdated kernel can mean security risks, especially the possibility of getting “rooted”. While we have tools that can monitor every aspect of a customer’s instance, we didn’t have an effective method to assure a customer was running a secure kernel. This new feature will help ensure the proper kernel is installed, and increase the security of customer instances.

We’ve rolled out this to not only help us with our managed VPS customers, but also it’s especially important to our managed cloud customers. In many cases, with our cloud based customers, we do not have access to update your kernel. If this is the case, the customer must update the kernel themselves.

Starting today, customers can check our monitoring system to see if their instance has an acceptable kernel installed. If not, the KERNEL service row will be colored red.

Managed VPS customers who may see an outdated kernel are not to worry. We typically update kernels within the week of an update. We will send out a general notice on supportem.com about this update, and will reboot your instance to update the kernel. No work is needed on your end.

Managed cloud customers if you need assistance with a kernel upgrade, please don’t hesitate contacting support.

Here are some recommendations to help optimize your WordPress for search engine optimization (SEO) but also for speed. Unfortunately, some of these options are only available when you go with a managed VPS, and not available in a shared hosting environment.

Replace Apache With LiteSpeed

We always recommend replacing Apache with another web server. Apache is great for compatibility, but the worst with web site performance. Dynamic applications like PHP based apps, don’t scale well with a high volume of traffic, and isn’t efficient with memory usage. We recommend using LiteSpeed because of it’s compatibility with Apache configuration files, it’s support of mod_rewrite entries, security and it’s performance with PHP.

LiteSpeed offers the performance of mod_php, but the security when using suPHP. Unlike a mod_php setup, the LiteSpeed web server runs PHP as your user/group. No more having to worry about the proper permissions for folder when uploading content through WordPress.

The added benefit when using LiteSpeed is compatibility with Apache configuration files. This means control panels like DirectAdmin are fully compatible with Apache, and we typically recommend this configuration. Yes LiteSpeed is a commercial application, but the cost is minimal and well worth the monthly fee.

Limit The Amount Of WordPress Plugins

Without knowing the plugins you use for your WordPress blog, it always makes sense to limit the amount of plugins you use. Unfortunately, not all plugins are created equal. Some plugins are complete resource hogs. Some plugins are coded poorly, and can affect the load time of every web page. Others insert unneeded HTML code on every web page, when only used on one page. If unsure, remove each plugin one by one and use a tool like Google Chrome Developer Tools to monitor the load time of your web pages.

Install a PHP Accelerator

PHP out of the box scales just ok. PHP is an interpreted language. Code written in PHP is converted into machine readable code (bytecode) and served through your web server. This process is time consuming, and adds overhead when every web page is loaded. It also makes your web site much more CPU bound, since PHP must compile the code every time.

Fortunately there is a solution to this issue. Install an opcode cache for PHP. There are many PHP accelerators available such as: eAcceleator, ionCube PHP Accelerator, XCache, and Zend Optimizer.

From our experience with compatibility, we recommend the PECL package APC. PHP accelerators are not perfect, and sometimes can cause issues with the created PHP code. From our experience APC seems to be the most reliable, but unfortunately isn’t the fastest. In tandem with a PHP accelerator, we also recommend installing a cache engine plugin within WordPress.

Use a Cache Engine

As mentioned previously, every web page with WordPress is generated dynamically with PHP scripts. This can slow down your web page generation time. We recommend not only using APC, but then adding a caching engine within WordPress. The two recommended WordPress plugins are:

• WP Super Cache
• W3 Total Cache

Each plugin has it’s advantages and disadvantages and should choose the one based upon your situation. We typically recommend using WP Super Cache first.

WP Super Cache
WP Super Cache converts your entire blog into static HTML. Static HTML always loads faster than dynamically generated PHP code, and can handle a much higher amount of web site hits per second. This plugin is perfect if all of your content is static and you do not have any dynamically generated content that changes often on a web page (ie stock quotes).

W3 Total Cache
Some plugins do not work with WP Super Cache. If this is your situation, we then recommend using W3 Total Cache. Unlike WP Super Cache which converts all of your WordPress blog into static content, W3 Total Cache uses various methods to cache code within PHP. This allows for better flexibility with the choice of plugins you use, but from our tests runs slightly slower then WP Super Cache. W3 Total Cache can be complicated to setup, but has many more options, is much more flexible, and can work with memcached.

Use the Genesis Framework

We recommend not using the default themes, or the many freely available themes for WordPress. Many WordPress themes are junk, or have hidden URLs that help the developer’s SEO. It is possible to hire a web developer to create a custom theme for your WordPress blog, but this can be time-consuming and costly.

A much cheaper alternative we recommend is the Genesis WordPress Framework by StudioPress. It’s a great WordPress theme to base your site on.

Fortunately if you aren’t a web designer, Genesis offers many child themes you can get your web site up and running in no time. Genesis not only allows for you to create quickly a professionally looking web site, but is optimized for SEO and quick loading. We recommend StudioPress Genesis Framework to all our customers.

MySQL Optimization

The SQL database MySQL is an integral part of WordPress. Since every page generated is stored within MySQL, it’s critical MySQL is optimized.

Depending upon the amount of memory available with your instance, and the size of your database it’s critical it’s all in memory and not using swap space. Using your instance disk swap tends to degrade performance very quickly. Therefore the configuration settings for your MySQL can be very dependent upon the amount of postings you have on your blog. Below is a general recommendation, but is not comprensive.

Recommended my.cnf

[mysqld]
local-infile=0
query_cache_size=84M
tmp_table_size=84M
max_heap_table_size=64M
thread_cache_size=4
table_cache=164


By default WordPress installs the database tables as MyISAM storage type. While this is ok for a small web site, this can cause problems with a site that has many reads and writes to it’s database tables. The MyISAM format performs table locking when updates to the table occur. This means if another process needs to write to the same database table, it must wait until the other process completes. This can cause delays in web page generation. MyISAM is really just a glorified flat file database format. Read for fast reads, but poor with most other database operations.

The table format InnoDB on the other hand supports row locking, which means multiple reads and writes can use the same table concurrently. This prevents delay in page generation. Not every table within WordPress must be converted though. Some tables can perform better with MyISAM since they are mainly read only.

From our experience, tables wp_options and wp_comments should be converted to InnoDB.

Use a CDN

Your web server is busy generating web pages. From your blog it also has to serve videos, images, stylesheets, Javascript, and other various static content. This increases the contention for your web server.

If you have a web site visitor coming from California, but your server is located in New York, their request must travel across the country. Latency, or the time it takes to across the country, is at minimum 100ms. Add this delay for the many requests needed to generate a complete web page, it can add easly 2-3 seconds to the load time of your web site.

Using a Content Delivery Network, or CDN, is the best option to speed up your static content. CDNs have multiple locations around the world to store your content. By pushing your content to a CDN, the web site visitor will get access to your content from the nearest location to them. This means decreased time to load your static content. The added benefit is your web server, which is now just focused on generating dynamic web pages, no longer has the additional strain of serving static files.

We recommend using Amazon’s Cloudfront service, but there are many other CDN providers available.

Pick An Outstanding Hosting Provider

The last item, but definitely not least option affecting the performance of your WordPress blog, is your choice of hosting provider. If using shared hosting, many of the options mentioned above are limited by the expertise and flexibility of your hosting provider. Some may have options to help with the performance of your blog, but most need to optimize for all their customers in a cookie cutter fashion.

Unfortunately it is also the dirty little secret many overload their shared hosting servers to ensure a profit. Services that are unlimited does not exist in the real world. They must limit something, even though they are offering “unlimited” disk space and bandwidth. Usually with dynamically generated web sites like WordPress blogs, the limiting factor is the amount of CPU power available. This means your site loads slowly, and is the primary factor to your site’s lag. In some cases the hosting provider will suspend your account because the amount of CPU used by your web site. If your web site is another other than a personal blog, you need to migrate to a VPS.

Virtual Private Servers (VPS) act just like a real server – you have your own dedicated CPU, memory, and disk space. Though compared to a dedicated server, a VPS is at the fraction of the cost.

Unfortunately many VPSes are unmanaged. This means you are responsible for everything: backups, security, software updates, and monitoring. Many people don’t have the time, or expertise to manage their own server.

For a little promotion of our services, HostCube offers a managed VPS or managed cloud services. We work within our own cloud, or with third party cloud providers like Amazon’s EC2.

Everything is done for you, and is no different than a shared hosting account. It service is still a black box to you, and you just have to focus on your blog.

The amount of time saved by hiring an hosting firm who’s an expert in optimizing WordPress is well worth the cost. While we have our hosting services, we also do consulting for customers who need help with their existing WordPress setup.

For the details, please read the release notes:

• CentOS 6.0 Release Notes
• CentOS 6.1 Release Notes
• CentOS 6.2 Release Notes

http://wiki.centos.org/Manuals/ReleaseNotes/CentOS5.7/

Customers who are currently have a CentOS 5 instance has been automatically upgraded.

http://wiki.centos.org/Manuals/ReleaseNotes/CentOS5.6/

For more information see our managed VPS web page.

Long time no post.  We had a Twitter account: @empoweringmedia for over a year, but did not formally announce this.

Stay tuned for more updates.

http://broadcast.oreilly.com/2008/12/why-i-dont-like-cloud-auto-scaling.html

The results have been interesting to say the least. For instance, let’s look at the rough costs of having your own internal server for one year:

• Hardware: $1500 Server, upgraded/replaced every 3 years = $500/yr
• Electricity: 165 watts continuous @ .14 per kWh = $202.50/yr
• Daily off site backups stored a minimum of 30 days: $800/yr

So, at this point we’re looking at approximately $1400 per year without even factoring in costs for:

• Administration
• T1 or better connection
• Infrastructure – cooling, battery backup, firewall, and physical space to house the equipment

For argument’s sake, we’ll say that you already have a T1 connection ($450+/month), but even if so, we still need to add on the costs of an administrator. Even if we factor in very affordable system administration services ($35/hr) and estimate only 2 hours of SysAdmin time per week, we’re looking at another $3640 per year, bringing our total costs to over $5000. These estimations are on the very low-end and in reality will usually cost much more.

Considering that HostCube VPS configurations start at $960/year and top out at $2400/year for the XL plan, we’re thinking you’d need an interesting set of requirements before it would make sense to seriously consider in-house hosting over a VPS. This is especially true as the VPS configuration would significantly outperform the in-house hosting configuration noted above.

In the weeks to come we’ll be comparing VPS to dedicated hosted servers as well as trying to peer through some of the fog surrounding “the cloud.”

BTW If you’re interested in the differences between managed VPS and shared hosting, check out Shared hosting vs. managed VPS. When to upgrade?

I’m not using this as scare tactic or as a reason to use our proactively managed virtual private servers, but as a fact from what we see every day. Unfortunately, the number of compromised servers we see hitting our network every day is staggering. For example, it’s not uncommon for us to block 50-100 new unique IP addresses daily for SSH or FTP brute force attempts. That’s just on a slow day. These are not Windows based desktops mind you, but Unix, Linux, or Windows based servers. You know, the servers that you would think are managed by system administration professionals.

Unfortunately it’s all too common for a web developer or designer think that buying a $99/month unmanaged dedicated server is all that is needed. Set it and forget it! Security is a continuous and never-ending process. Also many system administrators do not configure a server properly or too overtaxed putting out other fires. Even if you are not a HostCube customer, I strongly recommend hiring a qualified person or company to keep your server secure.

According to a Verizon Business study, the most common notification method of a security compromise is by a third party. So not only is the server software kept out of date, the software configuration is probably kept to defaults, but no monitoring exists when a comprise occurs. In this all too common case you might as well give away the keys and let the hacker do what they want with your setup!

Server security is about deploying layers; many different methods of proactive, reactive and defensive measures to protect from getting compromised. To give a high level view, here are the precautions we employ with managed virtual servers and managed servers:

• Remove unneeded services
• Employ best practices with software configuration (server hardening)
• Automatic deployment of software updates
• Hardware and local firewalls
• Remote log storage
• Backups (we can be used a method to audit a server)
• Scanning for common Rootkits
• Host Integrity Monitoring (HIM) to detect changes in operating system files
• Unusual port and services detection
• Monitoring services for availability and trends
• Autoblock SSH/FTP attacks
• Proper file and service permissions (sandboxing)

Monitoring log files and detecting server changes are some of the additional security measures we employ. We’ve just recently added Rootkit Hunter and Osiris host integrity monitoring to our list. Both tools monitor and detect attempted rootkits and any modifications to server setup. This additional monitoring makes our services fully PCI DSS compliant and just some of the features we perform to make sure a customer’s account is secure. You can rest assured knowing we are monitoring your managed virtual server very closely.

I won’t go into all of our security measures, nor details, but I will say the above list should be at least your security baseline. When choosing a provider, ask if they are really ‘managed’ and are performing these tasks for you. If not, you’ll definitely want to make sure you are addressing all these areas on your own.

When evaluating the different virtualization technologies we had very specific requirements. We wanted virtualization technology that allowed for:

• dedicated server like isolation.
• customizations to the installed operating system (i.e. kernel, iptables, etc.) just like a dedicated server.
• proven deployment.
• cannot oversell services (ensuring a specific level of quality of service).
• complete separation of each operating system installation.

The differences with Virtuozzo and Xen are:

• fixed memory and disk definitions.
• custom kernels.
• firewall configuration.
• isolation.

For our purposes Xen acts, breaths, and looks like a dedicated server.

Fixed Memory and Disk Definitions

In Xen’s current form, memory cannot be oversold. If the node only has 16 GB of ram, it means only 16 GB of ram can be allocated to all VPS instances. Virtuozzo offers bustable memory, whereas Xen has hard, fixed caps. Burstable memory is great if you have control over all of the VPSes (everyone is friendly), but when you have a diverse environment, we prefer hard memory caps (you’re guaranteed by the technology that you actually get what you pay for).

With Xen, like a real server, you get a specific amount of memory and swap space. This allows the operating system handle memory and swap to disk as needed. Memory on Virtuozzo is a much muddled situation. To the VPS instance, it all appears to be memory, when in fact it’s not. With Virtuozzo, It’s not uncommon for services to just die because no memory is available. Depending upon the provider, memory can be oversold and performance is then no better than shared hosting setups. Currently with Xen there is no way for us to do this and therefore you know that what you pay for is exactly what you are getting.

Firewall Configuration

Firewall configuration with Virtuozzo is very limited and you do not have access to the full iptables setup. This means there are many hoops you must go through to firewall your VPS instance. Most Virtuozzo providers configure their setup to have the node control the firewall configuration.

Custom Kernel

Xen allows for custom kernels. Each VPS instance can be unique. Need a special module, or custom kernel? With Virtuozzo, since the virtualization uses the same kernel for all VPS instances, you can’t do this. With Xen, each virtual server has a separate kernel and allows for increased security.

Isolation

This is an area that is typically more anecdotal than “hard facts,” and is always up for debate by both sides. While the Virtuozzo crowd states their virtualization technology performs better, the tests I’ve seen are not real world situations. The simple fact is that I’ve never heard of a Xen node getting overwhelmed and bringing down the other instances. With Virtuozzo we’ve seen this happen a lot and it is primarily because of two factors:

• Oversold memory and bustable memory.
• Hypervisor scheduler. While Xen’s scheduler options are much more limited, it works and works well.

In our case we’ve had many times where we’ve seen a specific VPS instance running at 100% CPU usage and high disk IO without the other VPS instances even missing a beat. With Virtuozzo it is assumed each VPS will play well with others and this simply is not always the case. The biggest limitation with Xen is it’s disk IO scheduler. The IO scheduler is somewhat simplistic, but does have the benefit that you can control unruly virtual servers and prevent them from affecting the other servers.

Summary

If you are looking for “dedicated-like” performance, customization, isolation and security, Xen is a perfect fit. With our managed virtual servers, it’s like getting a dedicated server for a fraction of the cost!

Also keep in mind that while we currently use Xen we are not married to it. We are always looking at the other virtualization technologies. With the recent announcement from Red Hat, KVM definitely has a chance. The way our cloud infrastructure is designed, the VM manager really doesn’t matter. In the end, what matters to us and our customers is running what works and is stable, secure and provides the best isolation. For now, Xen is that perfect fit, while Virtuozzo is simply not as robust.

The primary reasons for using a VPS:

• Dedicated resources (Quality of Service).
• More control of installed software.
• More secure.
• Dedicated IP address.

With shared hosting the hosting provider has to make sure each customer’s hosting configuration performs well. The administrator might be able to perform proactive measures, but in many cases this isn’t possible. This is because each shared hosting account uses the same memory, CPU and disk space. This is similar to a noisy neighbor in a massive apartment building. All it takes is one bad tenant to affect the others. Each customer must share the same resources on a shared hosting server.

Dedicated Resources (Quality of Service)

With a VPS you are allocated a fixed amount of resources (just like a dedicated server) and these resources are dedicated to you. Each customer is separated at the operating system level and another customer cannot affect your VPS. This increases quality of service since you have a specific amount of memory, CPU and disk given to your account. This is also the main reason for the differences in price between VPS and shared hosting.

A common issue on many oversold shared hosting providers is that you are suspended once you use too much CPU, memory or disk space. On a VPS you will never get suspended for this reason. Also, shared hosting’s dirty little secret is that many dynamically generated web pages (i.e. blog, forum, CMS, E-commerce) are primarily CPU bound. On a massively oversold shared server there is only so much CPU to go around.  These providers put 500-600 accounts on each server and for this reason the performance of ALL clients on that server is affected.

Other Advantages of Choosing a VPS Solution Include:

More Control of Installed Software

With a VPS we can customize the software to the customer’s exact specifications. Since software such as apache, PHP, MySQL, etc are dedicated to just your account. Conversely, shared hosting is configured to please the majority and exceptions are not possible. In addition, if you need a service (otherwise known as a daemon) running or custom programming libraries, this is all possible with a VPS.

More Secure

Since each VPS is separated at the operating system level, each customer is running in its own memory, CPU and disk space. This prevents your account from getting compromised when another customer forgets to update their blog software to fix security risks that have been discovered – if hackers get into their shared hosting account they can quickly move horizontally into your account if you’re on the same shared server.

Dedicated IP address

This is important for any service that uses the source IP address for reputation purposes. This is extremely important with outbound E-mail (SMTP) and significantly decreases the chance of blockage because another customer sent out spam. To put this in perspective, we’ve seen cases where one misguided salesperson sending out less than 200 emails in an email blast has caused IP-based blocking of an entire IP (including all the responsible senders on that IP).

The HostCube Advantage

There’s one advantage that can sometimes be at credited to shared hosting: the provider manages all the system administration. Fortunately, this service is also provided when using a service like HostCube Managed Virtual Servers. The components of system administration include:

• Backups
• Service monitoring
• Security monitoring
• Software updates
• Software configuration

There’s no need to worry if your VPS is secure, and is your site running as our managed VPSes give you the best of both worlds: the ability to work as if in a shared hosting environment with the performance of a dedicated server.

http://wiki.centos.org/Manuals/ReleaseNotes/CentOS5.2/

I’ve created a new blog to discuss the technical side of our service and all things “tech”. The Customer Service blog will just post service interruptions, major software upgrades and new services. Technical details and other ramblings of mine will be posted to “The Dev Null Blog”

http://www.empoweringmedia.com/blog/

The first post “Going from a dedicated server to a VPS. Am I downgrading?” discusses some of the advantages of using VPS technology.

Some future blog postings:

• Puppet. What is it and how we use it to manage our computing “cloud”
• Offering root with managed hosting
• When to upgrade from shared hosting to dedicated

I hope you enjoy it.

More details can be found at our customer service blog:

http://www.supportem.com/blog/article/167

The revised plans are as follows:

• Small – 10 GB disk space, 100 GB monthly bandwidth, 20 GB of backup space
• Medium – 20 GB disk space, 200 GB monthly bandwidth, 40 GB of backup space
• Large – 40 GB disk space, 400 GB monthly bandwidth, 80 GB of backup space
• X-Large – 80 GB disk space, 800 GB monthly bandwidth, 160 GB of backup space

Any existing customer who wants their plan upgraded to match the new allocations please create a ticket with support. Downtime for the upgrade is usually no more than 10 min.