Using Flash-Based Storage Without Compromising Transactions (or Performance)

Using Flash-Based Storage Without Compromising Transactions (or Performance)

Everyone wants fast write access, high performance, a smaller footprint, green IT solutions, and the lowest possible total cost of ownership (TCO). For these reasons, many organizations use solid state drives (SSDs) or other forms of flash memory instead of the traditional hard disk drives (HDDs).

Flash-based storage is designed to provide extremely high volumes of input/output (I/O) with very low latency. The typical use cases for flash-based storage are virtual desktop infrastructures (VDIs), which require high I/O speeds when booting up, and databases, which need fast read access.

The challenge I see with many organizations is that flash-based storage performance is hampered by synchronous replication. With synchronous replication, each write operation must be propagated to the remote site, which imposes additional latency on each write operation. This propagation delay affects any type of storage system, especially flash-based.

So, how can organizations benefit from flash technology and still maintain the goal of zero transaction loss (i.e., recovery point objective (RPO)=0)? In many cases, the business side of an organization cannot risk losing any transaction and therefore compromises on performance.

Several Axxana customers use Oracle’s high-performance Exadata solution to run Oracle databases. This combined compute and storage system is similar to flash-based storage, but the latency challenge is even greater because Exadata uses Oracle Data Guard for replication. The Oracle Data Guard replication solution uses IP networks that, in many cases, impose greater network latency than dedicated fiber / dense wavelength division multiplexing (DWDM) networks (used in the case of storage-based replication). IP networks also tend to drop connections more often and suffer from network congestion, thereby interfering with the implementation of synchronous replication.

Assessing Potential Transaction Loss

Many of our Oracle Data Guard customers have been very surprised to learn how many transactions are actually at risk! Oracle Enterprise Manager Cloud Control and other monitoring tools provide statistics of this risk in terms of megabytes or seconds (lag in seconds), but the more important question is how many actual transactions may be lost in case of a disaster.

Axxana developed a special monitoring tool specifically for this purpose: The Axxana Statistic Lag Analyzer tool. This free tool collects statistical information from the Oracle databases and presents it in easily readable graphs that display both Transport and Apply Lags in MBs, number of transactions, and seconds. The tool is SQL-based and is very easy to install and configure. The following graphs present the tool’s output:


The Axxana Statistic Lag Analyzer tool allows database or disaster recovery managers to measure and assess the real risk that their systems may be exposed to.


Picture1 LAG Seconds










Picture1 LAG Transactions










Picture1 LAG MBs












Achieving High Performance and Zero Data Loss

So what does Axxana offer to IT organizations that want the best possible performance and zero risk for data loss?

The Axxana Phoenix System for Oracle allows organizations to replicate with any asynchronous replication solution, while providing RPO=0 as if the data were protected by synchronous replication!

Because the Phoenix System for Oracle is installed right next to the organization’s primary databases using FC/IB/10GigE connectivity and is based on flash technology, it allows protection of the Oracle REDO, Archive, and Control files in a synchronous manner without adding latency. Organizations benefit from not only the best performance, but also zero transaction loss (RPO=0). In addition, the Axxana Phoenix System for Oracle can be configured with up to 3.2TB of protected Lag in a single system; this configuration provides the perfect protection against wide area network (WAN) link failures or network congestion by protecting the Archive files.

Sounds interesting?

Please watch the following video, which explains how our new and innovative solution can help organizations achieve both high performance and zero transaction loss—without using multiple data centers: