Mythics Blog

The Power of ZS3 Amplified by DTrace

Posted on August 12, 2014 by Erik Benner

Tags: Oracle Storage Appliance

The infinite appetite for data from our applications is a never ending challenge to the IT staff. Not only do we need to keep feeding the monster, but we also have to craft new innovative ways (like HCC database storage) to feed the beast, while also managing more complex environments at the same time. This is further complicated by the trend to consolidate multiple systems into private clouds to improve the ROI on hardware and software. What is a storage admin to do?

The ZS3 series from Oracle allows us to consolidate storage arrays in ways that boggle the mind. Not only can we mix file based workloads, like NFS, CIFS, FTP and even WEBDav on a single array, but we can also host our block storage needs on the same array. This allows a single array to easily do the work of multiple legacy arrays, while at the same time providing new space saving technologies to the application, like-array based compression and Hybrid Columnar Compression for our databases. We can even use technologies like Snapshots to backup and save space for our databases, or leverage the deduplication technology for our data. Combine all of this into one array with an aggregate throughput of 16,212.66 SPC-2 MBPS™ with SPC-2 Price Performance™ of $12.08 (5x savings over other arrays) and you quickly understand why the ZS3 array is the industry leader in price/performance.

Managing all of these storage needs quickly becomes a bottleneck for the storage admin, unless you use the ZS3 array. The ZS3 software is from the same folks that invented NFS, and includes a user friendly interface to DTrace. DTrace is a dynamic kernel-tracing framework invented by Sun Microsystems. This real-time ability allows the ZS3 admin to monitor and troubleshoot the array in on-the-fly.

Let’s take a quick look at the array management interface, and how DTrace can be used to troubleshoot a common performance issue.

When you first log into a ZS3 array, you see the status page. This screen allows the ZS3 admin to quickly identify issues and overall array health and performance.

Let’s walk through the different sections to better understand what we are seeing.

The Usage section shows us how much space is consumed in each disk pool. We also can monitor the memory usage of the controller. One ability that stands out on the ZS3 array is how it uses memory to augment the SSD and traditional disks on the array. The ability to keep hot-blocks in RAM provides signifigant performance advantages. Combined with the HSP technology, and you have an array that will outperform systems costing twice as much.

 

 

 

 

 

Under Services we can see what abilities are enabled on this array. From here we can see that this array is configured to support a variety of protocols.

 One interesting use of the array is to eliminate the need for tradition file servers on your network. You can integrate the ZS3 into your network, providing secure access to file shares just like a file server would. At the same time integrate the shares with your antivirus engine to ensure that all the files are infection free.

Another unique feature is provides the same file system access by a variety of protocols simultaneously. This allows a share to be accessed via NFS, SMB, HTTP and even FTP at the same time. This is very useful for multimedia applications, like video processing.

 

The Hardware section provides a quick look at the array hardware, identifying faulted components via the FMA system. The Oracle Fault Management Architecture (FMA) automatically detects and diagnoses underlying system problems and automatically responds by taking faulty components offline. This significantly reduces the time storage admins need to spend troubleshooting faults.

 

The main screen shows the critical services and resources for an overall health-check. This enables the admin to see in a real-time resources like CPU, disk, and network activity, while at the same time looking at protocol usage.

Each metric has a simple to understand icon, that shows the relative health of the resource. From Sunny to Catageory 5 huricanne. This is a very user-friendly way to communicate resource health.

 

 

 

 

In the main screen we also have the ability to drill down into a particular resource. This enables the admin to quickly identify what is consuming array resources. When using file shares, we can drill down to identify the individual file, and the host that is generating the traffic!

Let’s look at how this is done. First, I will use the dd command on one of the servers to generate some additional NFS traffic.

From the main status page, we see a spike in NFSv4 traffic.

We can click on the graph to start using DTrace Analytics to identify the file and server!

The first graph will show the NFSv4 traffic, and you can see the ops per second increase. Click on the drill icon to drill down for more information.

For this example we will select by client.

The client drill down will show all the clients using NFSv4, with the number of ops. We can select each client to highlight its number on the graph. For this example, we highlighted the 192.168.200.32 host.

Now to identify the files for this host, select it, and click the drill down button again and select by file name.

This will now show all files used by the host, with the number of ops per file. We can clearly see that the file “big.file” is consuming the majority of the operations, and that the traffic is generated from the 192.168.200.32 host. I guess I need to stop the dd command now.

As your drill down into a problem, you will build a worksheet, and it will look similar to this.

If need be, we can save the worksheet for future use.

This entire DTrace session takes only seconds to perform, enabling real-time troubleshooting if performance issues occur. No special diagnostic files need to be generated and sent to a vendor for lengthy analysis. This is only one of the hundreds of metrics that can be analyzed real-time on the ZS3 series. We can also verify storage latency, RAM usage and even CPU process on the array. 

Oracle ZS3 Storage Appliance delivers enterprise-class storage that reduces your overall storage expense. The ZS3 family of arrays offers a robust data storage platform that complements the high-throughput requirements of Oracle engineered systems, like the Oracle Exadata or Oracle SuperCluster servers. As the only storage platform engineered with your database in mind, the ZS3 Appliance is the best storage system for your applications, providing cost benefits for customers requiring a mix of file and block storage for your  enterprise applications, virtualization, storage consolidation, test and development, business intelligence/data warehousing, and data protection.

Comments

  • ! No comments yet

Leave a Comment